/*
- * Copyright (C) 1994-1995, Index Data I/S
+ * Copyright (C) 1994-1999, Index Data
* All rights reserved.
* Sebastian Hammer, Adam Dickmeiss
*
* $Log: kinput.c,v $
- * Revision 1.15 1996-03-21 14:50:09 adam
+ * Revision 1.31 1999-02-02 14:50:56 adam
+ * Updated WIN32 code specific sections. Changed header.
+ *
+ * Revision 1.30 1998/10/28 10:53:57 adam
+ * Added type cast to prevent warning.
+ *
+ * Revision 1.29 1998/06/11 15:41:39 adam
+ * Minor changes.
+ *
+ * Revision 1.28 1998/03/05 08:45:12 adam
+ * New result set model and modular ranking system. Moved towards
+ * descent server API. System information stored as "SGML" records.
+ *
+ * Revision 1.27 1998/02/17 10:32:52 adam
+ * Fixed bug: binary files weren't opened with flag b on NT.
+ *
+ * Revision 1.26 1998/01/29 13:39:13 adam
+ * Compress ISAM is default.
+ *
+ * Revision 1.25 1997/09/17 12:19:14 adam
+ * Zebra version corresponds to YAZ version 1.4.
+ * Changed Zebra server so that it doesn't depend on global common_resource.
+ *
+ * Revision 1.24 1997/09/09 13:38:07 adam
+ * Partial port to WIN95/NT.
+ *
+ * Revision 1.23 1997/09/04 13:57:39 adam
+ * Added O_BINARY for open calls.
+ *
+ * Revision 1.22 1997/02/12 20:39:45 adam
+ * Implemented options -f <n> that limits the log to the first <n>
+ * records.
+ * Changed some log messages also.
+ *
+ * Revision 1.21 1996/11/08 11:10:23 adam
+ * Buffers used during file match got bigger.
+ * Compressed ISAM support everywhere.
+ * Bug fixes regarding masking characters in queries.
+ * Redesigned Regexp-2 queries.
+ *
+ * Revision 1.20 1996/11/01 08:58:41 adam
+ * Interface to isamc system now includes update and delete.
+ *
+ * Revision 1.19 1996/10/29 14:09:46 adam
+ * Use of cisam system - enabled if setting isamc is 1.
+ *
+ * Revision 1.18 1996/06/04 10:18:59 adam
+ * Minor changes - removed include of ctype.h.
+ *
+ * Revision 1.17 1996/05/14 15:47:07 adam
+ * Cleanup of various buffer size entities.
+ *
+ * Revision 1.16 1996/04/09 10:05:20 adam
+ * Bug fix: prev_name buffer possibly too small; allocated in key_file_init.
+ *
+ * Revision 1.15 1996/03/21 14:50:09 adam
* File update uses modify-time instead of change-time.
*
* Revision 1.14 1996/02/07 14:06:37 adam
*/
#include <fcntl.h>
+#ifdef WIN32
+#include <io.h>
+#else
#include <unistd.h>
+#endif
#include <stdlib.h>
#include <string.h>
#include <stdio.h>
-#include <ctype.h>
#include <assert.h>
#include "index.h"
#define KEY_SIZE (1+sizeof(struct it_key))
-#define INP_NAME_MAX 8192
+#define INP_NAME_MAX 768
#define INP_BUF_START 60000
#define INP_BUF_ADD 400000
void getFnameTmp (char *fname, int no)
{
- sprintf (fname, TEMP_FNAME, no);
+ const char *pre;
+
+ pre = res_get_def (common_resource, "keyTmpDir", ".");
+ sprintf (fname, "%s/key%d.tmp", pre, no);
}
void key_file_chunk_read (struct key_file *f)
{
- int nr = 0, r, fd;
- char fname[256];
+ int nr = 0, r = 0, fd;
+ char fname[1024];
getFnameTmp (fname, f->no);
- fd = open (fname, O_RDONLY);
+ fd = open (fname, O_BINARY|O_RDONLY);
if (fd == -1)
{
logf (LOG_FATAL|LOG_ERRNO, "cannot open %s", fname);
f->length = 0;
f->readHandler = NULL;
f->buf = xmalloc (f->chunk);
- f->prev_name = xmalloc (256);
+ f->prev_name = xmalloc (INP_NAME_MAX);
*f->prev_name = '\0';
key_file_chunk_read (f);
return f;
int heapnum;
int *ptr;
int (*cmp)(const void *p1, const void *p2);
+ Dict dict;
+ ISAM isam;
+ ISAMC isamc;
};
struct heap_info *key_heap_init (int nkeys,
for (i = 0; i<= nkeys; i++)
{
hi->ptr[i] = i;
- hi->info.buf[i] = xmalloc (512);
+ hi->info.buf[i] = xmalloc (INP_NAME_MAX);
}
return hi;
}
static int heap_read_one (struct heap_info *hi, char *name, char *key)
{
int n, r;
- char rbuf[512];
+ char rbuf[INP_NAME_MAX];
struct key_file *kf;
if (!hi->heapnum)
return 1;
}
-int heap_inp (Dict dict, ISAM isam, struct heap_info *hi)
+struct heap_cread_info {
+ char prev_name[INP_NAME_MAX];
+ char cur_name[INP_NAME_MAX];
+ char *key;
+ struct heap_info *hi;
+ int mode;
+ int more;
+};
+
+int heap_cread_item (void *vp, char **dst, int *insertMode)
+{
+ struct heap_cread_info *p = vp;
+ struct heap_info *hi = p->hi;
+
+ if (p->mode == 1)
+ {
+ *insertMode = p->key[0];
+ memcpy (*dst, p->key+1, sizeof(struct it_key));
+ (*dst) += sizeof(struct it_key);
+ p->mode = 2;
+ return 1;
+ }
+ strcpy (p->prev_name, p->cur_name);
+ if (!(p->more = heap_read_one (hi, p->cur_name, p->key)))
+ return 0;
+ if (*p->cur_name && strcmp (p->cur_name, p->prev_name))
+ {
+ p->mode = 1;
+ return 0;
+ }
+ *insertMode = p->key[0];
+ memcpy (*dst, p->key+1, sizeof(struct it_key));
+ (*dst) += sizeof(struct it_key);
+ return 1;
+}
+
+int heap_inpc (struct heap_info *hi)
+{
+ struct heap_cread_info hci;
+ ISAMC_I isamc_i = xmalloc (sizeof(*isamc_i));
+
+ hci.key = xmalloc (KEY_SIZE);
+ hci.mode = 1;
+ hci.hi = hi;
+ hci.more = heap_read_one (hi, hci.cur_name, hci.key);
+
+ isamc_i->clientData = &hci;
+ isamc_i->read_item = heap_cread_item;
+
+ while (hci.more)
+ {
+ char this_name[INP_NAME_MAX];
+ ISAMC_P isamc_p, isamc_p2;
+ char *dict_info;
+
+ strcpy (this_name, hci.cur_name);
+ assert (hci.cur_name[1]);
+ no_diffs++;
+ if ((dict_info = dict_lookup (hi->dict, hci.cur_name)))
+ {
+ memcpy (&isamc_p, dict_info+1, sizeof(ISAMC_P));
+ isamc_p2 = isc_merge (hi->isamc, isamc_p, isamc_i);
+ if (!isamc_p2)
+ {
+ no_deletions++;
+ if (!dict_delete (hi->dict, this_name))
+ abort();
+ }
+ else
+ {
+ no_updates++;
+ if (isamc_p2 != isamc_p)
+ dict_insert (hi->dict, this_name,
+ sizeof(ISAMC_P), &isamc_p2);
+ }
+ }
+ else
+ {
+ isamc_p = isc_merge (hi->isamc, 0, isamc_i);
+ no_insertions++;
+ dict_insert (hi->dict, this_name, sizeof(ISAMC_P), &isamc_p);
+ }
+ }
+ xfree (isamc_i);
+ return 0;
+}
+
+int heap_inp (struct heap_info *hi)
{
char *info;
- char next_name[INP_NAME_MAX+1];
- char cur_name[INP_NAME_MAX+1];
+ char next_name[INP_NAME_MAX];
+ char cur_name[INP_NAME_MAX];
int key_buf_size = INP_BUF_START;
int key_buf_ptr;
char *next_key;
int more;
next_key = xmalloc (KEY_SIZE);
- key_buf = xmalloc (key_buf_size * (KEY_SIZE));
+ key_buf = xmalloc (key_buf_size);
more = heap_read_one (hi, cur_name, key_buf);
while (more) /* EOF ? */
{
break;
memcpy (key_buf + key_buf_ptr, next_key, KEY_SIZE);
key_buf_ptr += KEY_SIZE;
- if (key_buf_ptr+KEY_SIZE >= key_buf_size)
+ if (key_buf_ptr+(int) KEY_SIZE >= key_buf_size)
{
char *new_key_buf;
new_key_buf = xmalloc (key_buf_size + INP_BUF_ADD);
}
no_diffs++;
nmemb = key_buf_ptr / KEY_SIZE;
- assert (nmemb*KEY_SIZE == key_buf_ptr);
- if ((info = dict_lookup (dict, cur_name)))
+ assert (nmemb * (int) KEY_SIZE == key_buf_ptr);
+ if ((info = dict_lookup (hi->dict, cur_name)))
{
ISAM_P isam_p, isam_p2;
- logf (LOG_DEBUG, "updating %s", cur_name);
memcpy (&isam_p, info+1, sizeof(ISAM_P));
- isam_p2 = is_merge (isam, isam_p, nmemb, key_buf);
+ isam_p2 = is_merge (hi->isam, isam_p, nmemb, key_buf);
if (!isam_p2)
{
no_deletions++;
- if (!dict_delete (dict, cur_name))
+ if (!dict_delete (hi->dict, cur_name))
abort ();
}
else
{
no_updates++;
if (isam_p2 != isam_p)
- dict_insert (dict, cur_name, sizeof(ISAM_P), &isam_p2);
+ dict_insert (hi->dict, cur_name, sizeof(ISAM_P), &isam_p2);
}
}
else
{
ISAM_P isam_p;
- logf (LOG_DEBUG, "inserting %s", cur_name);
no_insertions++;
- isam_p = is_merge (isam, 0, nmemb, key_buf);
- dict_insert (dict, cur_name, sizeof(ISAM_P), &isam_p);
+ isam_p = is_merge (hi->isam, 0, nmemb, key_buf);
+ dict_insert (hi->dict, cur_name, sizeof(ISAM_P), &isam_p);
}
memcpy (key_buf, next_key, KEY_SIZE);
strcpy (cur_name, next_name);
if (now >= p->lastTime+10)
{
p->lastTime = now;
- remaining = (now - p->startTime)*
- ((double) p->totalBytes/p->totalOffset - 1.0);
+ remaining = (time_t) ((now - p->startTime)*
+ ((double) p->totalBytes/p->totalOffset - 1.0));
if (remaining <= 130)
logf (LOG_LOG, "Merge %2.1f%% completed; %ld seconds remaining",
(100.0*p->totalOffset) / p->totalBytes, (long) remaining);
p->totalOffset += keyp->buf_size;
}
-void key_input (const char *dict_fname, const char *isam_fname,
- int nkeys, int cache)
+#ifndef R_OK
+#define R_OK 4
+#endif
+
+void key_input (BFiles bfs, int nkeys, int cache)
{
Dict dict;
- ISAM isam;
+ ISAM isam = NULL;
+ ISAMC isamc = NULL;
struct key_file **kf;
char rbuf[1024];
int i, r;
struct heap_info *hi;
struct progressInfo progressInfo;
- dict = dict_open (dict_fname, cache, 1);
+ if (nkeys < 0)
+ {
+ char fname[1024];
+ nkeys = 0;
+ while (1)
+ {
+ getFnameTmp (fname, nkeys+1);
+ if (access (fname, R_OK) == -1)
+ break;
+ nkeys++;
+ }
+ if (!nkeys)
+ return ;
+ }
+ dict = dict_open (bfs, FNAME_DICT, cache, 1);
if (!dict)
{
- logf (LOG_FATAL, "dict_open fail of `%s'", dict_fname);
+ logf (LOG_FATAL, "dict_open fail");
exit (1);
}
- isam = is_open (isam_fname, key_compare, 1, sizeof(struct it_key));
- if (!isam)
+ if (!res_get_match (common_resource, "isam", "i", NULL))
{
- logf (LOG_FATAL, "is_open fail of `%s'", isam_fname);
- exit (1);
+ isamc = isc_open (bfs,
+ FNAME_ISAMC, 1, key_isamc_m (common_resource));
+ if (!isamc)
+ {
+ logf (LOG_FATAL, "isc_open fail");
+ exit (1);
+ }
+ }
+ else
+ {
+ isam = is_open (bfs, FNAME_ISAM, key_compare, 1,
+ sizeof(struct it_key), common_resource);
+ if (!isam)
+ {
+ logf (LOG_FATAL, "is_open fail");
+ exit (1);
+ }
}
-
kf = xmalloc ((1+nkeys) * sizeof(*kf));
progressInfo.totalBytes = 0;
progressInfo.totalOffset = 0;
progressInfo.totalOffset += kf[i]->buf_size;
}
hi = key_heap_init (nkeys, key_qsort_compare);
+ hi->dict = dict;
+ hi->isam = isam;
+ hi->isamc = isamc;
+
for (i = 1; i<=nkeys; i++)
if ((r = key_file_read (kf[i], rbuf)))
key_heap_insert (hi, rbuf, r, kf[i]);
- heap_inp (dict, isam, hi);
-
+ if (isamc)
+ heap_inpc (hi);
+ else
+ heap_inp (hi);
dict_close (dict);
- is_close (isam);
-
+ if (isam)
+ is_close (isam);
+ if (isamc)
+ isc_close (isamc);
+
for (i = 1; i<=nkeys; i++)
{
getFnameTmp (rbuf, i);
projects / idzebra-moved-to-github.git / blobdiff