-/*
- * Copyright (C) 1994-2002, Index Data
- * All rights reserved.
- * Sebastian Hammer, Adam Dickmeiss
- *
- * $Id: extract.c,v 1.114 2002-04-04 14:14:13 adam Exp $
- */
+/* $Id: extract.c,v 1.126 2002-10-22 12:51:08 adam Exp $
+ Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
+ Index Data Aps
+
+This file is part of the Zebra server.
+
+Zebra is free software; you can redistribute it and/or modify it under
+the terms of the GNU General Public License as published by the Free
+Software Foundation; either version 2, or (at your option) any later
+version.
+
+Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
+WARRANTY; without even the implied warranty of MERCHANTABILITY or
+FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
+for more details.
+
+You should have received a copy of the GNU General Public License
+along with Zebra; see the file LICENSE.zebra. If not, write to the
+Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
+02111-1307, USA.
+*/
+
+
#include <stdio.h>
#include <assert.h>
#ifdef WIN32
#include <fcntl.h>
#include "index.h"
+#include <direntz.h>
#include <charmap.h>
#if _FILE_OFFSET_BITS == 64
#define PRINTF_OFF_T "%ld"
#endif
-static int records_inserted = 0;
-static int records_updated = 0;
-static int records_deleted = 0;
-static int records_processed = 0;
+#define USE_SHELLSORT 0
-static void logRecord (int showFlag)
+#if USE_SHELLSORT
+static void shellsort(void *ar, int r, size_t s,
+ int (*cmp)(const void *a, const void *b))
{
- if (!showFlag)
- ++records_processed;
- if (showFlag || !(records_processed % 1000))
+ char *a = ar;
+ char v[100];
+ int h, i, j, k;
+ static const int incs[16] = { 1391376, 463792, 198768, 86961, 33936,
+ 13776, 4592, 1968, 861, 336,
+ 112, 48, 21, 7, 3, 1 };
+ for ( k = 0; k < 16; k++)
+ for (h = incs[k], i = h; i < r; i++)
+ {
+ memcpy (v, a+s*i, s);
+ j = i;
+ while (j > h && (*cmp)(a + s*(j-h), v) > 0)
+ {
+ memcpy (a + s*j, a + s*(j-h), s);
+ j -= h;
+ }
+ memcpy (a+s*j, v, s);
+ }
+}
+#endif
+
+static void logRecord (ZebraHandle zh)
+{
+ ++zh->records_processed;
+ if (!(zh->records_processed % 1000))
{
logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
- records_processed, records_inserted, records_updated,
- records_deleted);
+ zh->records_processed, zh->records_inserted, zh->records_updated,
+ zh->records_deleted);
}
}
static void extract_init (struct recExtractCtrl *p, RecWord *w)
{
w->zebra_maps = p->zebra_maps;
- w->seqnos = p->seqno;
+ w->seqno = 1;
w->attrSet = VAL_BIB1;
w->attrUse = 1016;
w->reg_type = 'w';
struct recordGroup *rGroup;
};
-static void recordLogPreamble (int level, const char *msg, void *info)
-{
- struct recordLogInfo *p = (struct recordLogInfo *) info;
- FILE *outf = yaz_log_file ();
-
- if (level & LOG_LOG)
- return ;
- fprintf (outf, "File %s, offset %d, type %s\n",
- p->rGroup->recordType, p->recordOffset, p->fname);
- log_event_start (NULL, NULL);
-}
-
-
static int recordExtract (ZebraHandle zh,
SYSNO *sysno, const char *fname,
struct recordGroup *rGroup, int deleteFlag,
char *matchStr;
SYSNO sysnotmp;
Record rec;
- struct recordLogInfo logInfo;
off_t recordOffset = 0;
if (fi->fd != -1)
if (!rGroup->flagRw)
printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
-
- logInfo.fname = fname;
- logInfo.recordOffset = recordOffset;
- logInfo.rGroup = rGroup;
- log_event_start (recordLogPreamble, &logInfo);
+ if (rGroup->flagRw)
+ {
+ char msg[512];
+ sprintf (msg, "%s:" PRINTF_OFF_T , fname, recordOffset);
+ yaz_log_init_prefix2 (msg);
+ }
r = (*recType->extract)(clientData, &extractCtrl);
- log_event_start (NULL, NULL);
-
+ yaz_log_init_prefix2 (0);
if (r == RECCTRL_EXTRACT_EOF)
return 0;
- else if (r == RECCTRL_EXTRACT_ERROR)
+ else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
{
/* error occured during extraction ... */
if (rGroup->flagRw &&
- records_processed < rGroup->fileVerboseLimit)
+ zh->records_processed < rGroup->fileVerboseLimit)
{
logf (LOG_WARN, "fail %s %s " PRINTF_OFF_T, rGroup->recordType,
fname, recordOffset);
}
return 0;
}
+ else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
+ {
+ /* error occured during extraction ... */
+ if (rGroup->flagRw &&
+ zh->records_processed < rGroup->fileVerboseLimit)
+ {
+ logf (LOG_WARN, "no filter for %s %s "
+ PRINTF_OFF_T, rGroup->recordType,
+ fname, recordOffset);
+ }
+ return 0;
+ }
if (zh->reg->keys.buf_used == 0)
{
/* the extraction process returned no information - the record
logf (LOG_WARN, "cannot delete record above (seems new)");
return 1;
}
- if (records_processed < rGroup->fileVerboseLimit)
+ if (zh->records_processed < rGroup->fileVerboseLimit)
logf (LOG_LOG, "add %s %s " PRINTF_OFF_T, rGroup->recordType,
fname, recordOffset);
rec = rec_new (zh->reg->records);
extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
- records_inserted++;
+ zh->records_inserted++;
}
else
{
if (recordAttr->runNumber ==
zebraExplain_runNumberIncrement (zh->reg->zei, 0))
{
- logf (LOG_LOG, "skipped %s %s " PRINTF_OFF_T, rGroup->recordType,
- fname, recordOffset);
+ yaz_log (LOG_LOG, "run number = %d", recordAttr->runNumber);
+ yaz_log (LOG_LOG, "skipped %s %s " PRINTF_OFF_T,
+ rGroup->recordType, fname, recordOffset);
extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
rec_rm (&rec);
- logRecord (0);
+ logRecord (zh);
return 1;
}
delkeys.buf_used = rec->size[recInfo_delKeys];
}
else
{
- if (records_processed < rGroup->fileVerboseLimit)
+ if (zh->records_processed < rGroup->fileVerboseLimit)
logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T,
rGroup->recordType, fname, recordOffset);
- records_deleted++;
+ zh->records_deleted++;
if (matchStr)
dict_delete (zh->reg->matchDict, matchStr);
rec_del (zh->reg->records, &rec);
}
rec_rm (&rec);
- logRecord (0);
+ logRecord (zh);
return 1;
}
else
}
else
{
- if (records_processed < rGroup->fileVerboseLimit)
+ if (zh->records_processed < rGroup->fileVerboseLimit)
logf (LOG_LOG, "update %s %s " PRINTF_OFF_T,
rGroup->recordType, fname, recordOffset);
extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
- records_updated++;
+ zh->records_updated++;
}
}
}
/* commit this record */
rec_put (zh->reg->records, &rec);
- logRecord (0);
+ logRecord (zh);
return 1;
}
}
if (!rGroup->recordType)
{
- if (records_processed < rGroup->fileVerboseLimit)
+ if (zh->records_processed < rGroup->fileVerboseLimit)
logf (LOG_LOG, "? %s", fname);
return 0;
}
fd = -1;
else
{
- if ((fd = open (fname, O_BINARY|O_RDONLY)) == -1)
+ char full_rep[1024];
+
+ if (zh->path_reg && !yaz_is_abspath (fname))
+ {
+ strcpy (full_rep, zh->path_reg);
+ strcat (full_rep, "/");
+ strcat (full_rep, fname);
+ }
+ else
+ strcpy (full_rep, fname);
+
+
+ if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
{
- logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
+ logf (LOG_WARN|LOG_ERRNO, "open %s", full_rep);
return 0;
}
}
if (r == RECCTRL_EXTRACT_EOF)
return 0;
- else if (r == RECCTRL_EXTRACT_ERROR)
+ else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
{
/* error occured during extraction ... */
-#if 1
- yaz_log (LOG_WARN, "extract error");
-#else
- if (rGroup->flagRw &&
- records_processed < rGroup->fileVerboseLimit)
- {
- logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
- fname, (long) recordOffset);
- }
-#endif
+ yaz_log (LOG_WARN, "extract error: generic");
+ return 0;
+ }
+ else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
+ {
+ /* error occured during extraction ... */
+ yaz_log (LOG_WARN, "extract error: no such filter");
return 0;
}
if (zh->reg->keys.buf_used == 0)
if (!zh->reg->key_buf)
{
- int mem = 8*1024*1024;
+ int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
+ if (mem <= 0)
+ {
+ logf(LOG_WARN, "Invalid memory setting, using default 8 MB");
+ mem= 1024*1024*8;
+ }
+ /* FIXME: That "8" should be in a default settings include */
+ /* not hard-coded here! -H */
zh->reg->key_buf = (char**) xmalloc (mem);
zh->reg->ptr_top = mem/sizeof(char*);
zh->reg->ptr_i = 0;
(zh->reg->key_file_no)++;
logf (LOG_LOG, "sorting section %d", (zh->reg->key_file_no));
#if !SORT_EXTRA
- qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i, sizeof(char*),
- key_qsort_compare);
+ qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
+ sizeof(char*), key_qsort_compare);
extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
if (!(outf = fopen (out_fname, "wb")))
cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
if (strcmp (cp, prevcp))
{
+ encode_key_flush ( &encode_info, outf);
encode_key_init (&encode_info);
encode_key_write (cp, &encode_info, outf);
prevcp = cp;
else
encode_key_write (cp + strlen(cp), &encode_info, outf);
}
+ encode_key_flush ( &encode_info, outf);
#else
qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
extract_get_fname_tmp (out_fname, key_file_no);
cp = key_buf[ptr_top-ptr_i];
encode_key_write (cp+key_y_len, &encode_info, outf);
}
+ encode_key_flush ( &encode_info, outf);
if (!i)
break;
prevcp = key_buf[ptr_top-ptr_i];
unsigned short attrUse;
int lead = 0;
int diff = 0;
- int *pseqno = &p->seqnos[p->reg_type];
+ int *pseqno = &p->seqno;
ZebraHandle zh = p->extractCtrl->handle;
ZebraExplainInfo zei = zh->reg->zei;
struct recKeys *keys = &zh->reg->keys;
-
+
if (keys->buf_used+1024 > keys->buf_max)
{
char *b;
dst += sizeof(*pseqno);
}
keys->buf_used = dst - keys->buf;
- if (*pseqno)
- (*pseqno)++;
}
static void extract_add_sort_string (RecWord *p, const char *string,
if (!i)
return;
extract_add_string (p, buf, i);
+ p->seqno++;
}
- (p->seqnos[p->reg_type])++; /* to separate this from next one */
}
static void extract_add_complete_field (RecWord *p)
while (sk)
{
struct sortKey *sk_next = sk->next;
- sortIdx_type (sortIdx, sk->attrUse);
- sortIdx_add (sortIdx, sk->string, sk->length);
+ if (cmd == 1)
+ {
+ /* insert/update: set it */
+ sortIdx_type (sortIdx, sk->attrUse);
+ sortIdx_add (sortIdx, sk->string, sk->length);
+ }
+ else if (cmd == 0)
+ { /* delete : zero it */
+ sortIdx_type (sortIdx, sk->attrUse);
+ sortIdx_add (sortIdx, "", 1);
+ }
xfree (sk->string);
xfree (sk);
sk = sk_next;
i->sysno = 0;
i->seqno = 0;
i->cmd = -1;
+ i->prevsys=0;
+ i->prevseq=0;
+ i->prevcmd=-1;
+ i->keylen=0;
}
char *encode_key_int (int d, char *bp)
return bp;
}
+#ifdef OLDENCODE
+/* this is the old encode_key_write
+ * may be deleted once we are confident that the new works
+ * HL 15-oct-2002
+ */
void encode_key_write (char *k, struct encode_info *i, FILE *outf)
{
struct it_key key;
}
}
+void encode_key_flush (struct encode_info *i, FILE *outf)
+{ /* dummy routine */
+}
+
+#else
+
+/* new encode_key_write
+ * The idea is to buffer one more key, and compare them
+ * If we are going to delete and insert the same key,
+ * we may as well not bother. Should make a difference in
+ * updates with small modifications (appending to a mbox)
+ */
+void encode_key_write (char *k, struct encode_info *i, FILE *outf)
+{
+ struct it_key key;
+ char *bp;
+
+ if (*k) /* first time for new key */
+ {
+ bp = i->buf;
+ while ((*bp++ = *k++))
+ ;
+ i->keylen= bp - i->buf -1;
+ }
+ else
+ {
+ bp=i->buf + i->keylen;
+ *bp++=0;
+ k++;
+ }
+
+ memcpy (&key, k+1, sizeof(struct it_key));
+ if (0==i->prevsys) /* no previous filter, fill up */
+ {
+ i->prevsys=key.sysno;
+ i->prevseq=key.seqno;
+ i->prevcmd=*k;
+ }
+ else if ( (i->prevsys==key.sysno) &&
+ (i->prevseq==key.seqno) &&
+ (i->prevcmd!=*k) )
+ { /* same numbers, diff cmd, they cancel out */
+ i->prevsys=0;
+ }
+ else
+ { /* different stuff, write previous, move buf */
+ bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
+ if (i->sysno != i->prevsys)
+ {
+ i->sysno = i->prevsys;
+ i->seqno = 0;
+ }
+ else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
+ {
+ return; /* ??? Filters some sort of duplicates away */
+ /* ??? Can this ever happen -H 15oct02 */
+ }
+ bp = encode_key_int (i->prevseq - i->seqno, bp);
+ i->seqno = i->prevseq;
+ i->cmd = i->prevcmd;
+ if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
+ {
+ logf (LOG_FATAL|LOG_ERRNO, "fwrite");
+ exit (1);
+ }
+ i->keylen=0; /* ok, it's written, forget it */
+ i->prevsys=key.sysno;
+ i->prevseq=key.seqno;
+ i->prevcmd=*k;
+ }
+#ifdef SKIPTHIS_OLDCODE
+ bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
+ if (i->sysno != key.sysno)
+ {
+ 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");
+ exit (1);
+ }
+ i->keylen=0; /* ok, it's written, forget it */
+#endif
+}
+
+void encode_key_flush (struct encode_info *i, FILE *outf)
+{ /* flush the last key from i */
+ char *bp =i->buf + i->keylen;
+ if (0==i->prevsys)
+ {
+ return; /* nothing to flush */
+ }
+ *bp++=0;
+ bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
+ if (i->sysno != i->prevsys)
+ {
+ i->sysno = i->prevsys;
+ i->seqno = 0;
+ }
+ else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
+ {
+ return; /* ??? Filters some sort of duplicates away */
+ /* ??? Can this ever happen -H 15oct02 */
+ }
+ bp = encode_key_int (i->prevseq - i->seqno, bp);
+ i->seqno = i->prevseq;
+ i->cmd = i->prevcmd;
+ if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
+ {
+ logf (LOG_FATAL|LOG_ERRNO, "fwrite");
+ exit (1);
+ }
+ i->keylen=0; /* ok, it's written, forget it */
+ i->prevsys=0; /* forget the values too */
+ i->prevseq=0;
+}
+#endif