/*
- * Copyright (C) 1994-1995, Index Data I/S
+ * Copyright (C) 1994-2002, Index Data
* All rights reserved.
* Sebastian Hammer, Adam Dickmeiss
*
- * $Log: rstemp.c,v $
- * Revision 1.10 1995-09-15 14:45:39 adam
- * Bug fixes.
- *
- * Revision 1.9 1995/09/15 09:20:42 adam
- * Bug fixes.
- *
- * Revision 1.8 1995/09/08 14:52:42 adam
- * Work on relevance feedback.
- *
- * Revision 1.7 1995/09/07 13:58:44 adam
- * New parameter: result-set file descriptor (RSFD) to support multiple
- * positions within the same result-set.
- * Boolean operators: and, or, not implemented.
- *
- * Revision 1.6 1995/09/06 16:11:56 adam
- * More work on boolean sets.
- *
- * Revision 1.5 1995/09/05 16:36:59 adam
- * Minor changes.
- *
- * Revision 1.4 1995/09/05 11:43:24 adam
- * Complete version of temporary sets. Not tested yet though.
- *
- * Revision 1.3 1995/09/04 15:20:40 adam
- * More work on temp sets. is_open member removed.
- *
- * Revision 1.2 1995/09/04 09:10:56 adam
- * Minor changes.
- *
- * Revision 1.1 1994/11/04 13:21:30 quinn
- * Working.
- *
+ * $Id: rstemp.c,v 1.30 2002-03-20 20:24:30 adam Exp $
*/
#include <fcntl.h>
#include <assert.h>
+#ifdef WIN32
+#include <io.h>
+#else
#include <unistd.h>
+#endif
+#include <string.h>
#include <sys/types.h>
#include <stdio.h>
-#include <alexutil.h>
+#include <zebrautl.h>
#include <rstemp.h>
-static rset_control *r_create(const struct rset_control *sel, void *parms);
-static RSFD r_open (rset_control *ct, int wflag);
+static void *r_create(RSET ct, const struct rset_control *sel, void *parms);
+static RSFD r_open (RSET ct, int flag);
static void r_close (RSFD rfd);
-static void r_delete (rset_control *ct);
+static void r_delete (RSET ct);
static void r_rewind (RSFD rfd);
-static int r_count (rset_control *ct);
-static int r_read (RSFD rfd, void *buf);
+static int r_count (RSET ct);
+static int r_read (RSFD rfd, void *buf, int *term_index);
static int r_write (RSFD rfd, const void *buf);
-static const rset_control control =
+static const struct rset_control control =
{
- "Temporary set",
- 0,
+ "temp",
r_create,
r_open,
r_close,
r_rewind,
r_count,
r_read,
- r_write
+ r_write,
};
-const rset_control *rset_kind_temp = &control;
+const struct rset_control *rset_kind_temp = &control;
struct rset_temp_info {
int fd;
size_t pos_buf; /* position of first byte in window */
size_t pos_border; /* position of last byte+1 in window */
int dirty; /* window is dirty */
+ int hits; /* no of hits */
+ char *temp_path;
+ int (*cmp)(const void *p1, const void *p2);
};
struct rset_temp_rfd {
struct rset_temp_info *info;
struct rset_temp_rfd *next;
+ int *countp;
+ void *buf;
};
-static struct rset_control *r_create(const struct rset_control *sel,
- void *parms)
+static void *r_create(RSET ct, const struct rset_control *sel, void *parms)
{
- rset_control *newct;
- rset_temp_parms *temp_parms = parms;
+ rset_temp_parms *temp_parms = (rset_temp_parms *) parms;
struct rset_temp_info *info;
-
- newct = xmalloc(sizeof(*newct));
- memcpy(newct, sel, sizeof(*sel));
- newct->buf = xmalloc (sizeof(struct rset_temp_info));
- info = newct->buf;
-
+
+ info = (struct rset_temp_info *) xmalloc (sizeof(struct rset_temp_info));
info->fd = -1;
info->fname = NULL;
info->key_size = temp_parms->key_size;
- info->buf_size = 2048;
- info->buf_mem = xmalloc (info->buf_size);
+ info->buf_size = 4096;
+ info->buf_mem = (char *) xmalloc (info->buf_size);
info->pos_cur = 0;
info->pos_end = 0;
info->pos_buf = 0;
info->dirty = 0;
+ info->hits = -1;
+ info->cmp = temp_parms->cmp;
+ if (!temp_parms->temp_path)
+ info->temp_path = NULL;
+ else
+ {
+ info->temp_path = (char *) xmalloc (strlen(temp_parms->temp_path)+1);
+ strcpy (info->temp_path, temp_parms->temp_path);
+ }
+ ct->no_rset_terms = 1;
+ ct->rset_terms = (RSET_TERM *) xmalloc (sizeof(*ct->rset_terms));
+ ct->rset_terms[0] = temp_parms->rset_term;
- return newct;
+ return info;
}
-static RSFD r_open (struct rset_control *ct, int wflag)
+static RSFD r_open (RSET ct, int flag)
{
- struct rset_temp_info *info = ct->buf;
+ struct rset_temp_info *info = (struct rset_temp_info *) ct->buf;
struct rset_temp_rfd *rfd;
assert (info->fd == -1);
if (info->fname)
{
- if (wflag)
- info->fd = open (info->fname, O_RDWR|O_CREAT, 0666);
+ if (flag & RSETF_WRITE)
+ info->fd = open (info->fname, O_BINARY|O_RDWR|O_CREAT, 0666);
else
- info->fd = open (info->fname, O_RDONLY);
+ info->fd = open (info->fname, O_BINARY|O_RDONLY);
if (info->fd == -1)
{
logf (LOG_FATAL|LOG_ERRNO, "open %s", info->fname);
exit (1);
}
}
- rfd = xmalloc (sizeof(*rfd));
+ rfd = (struct rset_temp_rfd *) xmalloc (sizeof(*rfd));
rfd->info = info;
r_rewind (rfd);
+
+ rfd->countp = &ct->rset_terms[0]->count;
+ rfd->buf = xmalloc (info->key_size);
+
return rfd;
}
if (!info->fname && mk)
{
- char *s = (char*) tempnam (NULL, "zrs");
+#if HAVE_MKSTEMP
+ char template[1024];
- info->fname = xmalloc (strlen(s)+1);
+ sprintf (template, "%s/zrsXXXXXX", info->temp_path);
+
+ info->fd = mkstemp (template);
+
+ if (info->fd == -1)
+ {
+ logf (LOG_FATAL|LOG_ERRNO, "mkstemp %s", template);
+ exit (1);
+ }
+ info->fname = (char *) xmalloc (strlen(template)+1);
+ strcpy (info->fname, template);
+#else
+ char *s = (char*) tempnam (info->temp_path, "zrs");
+ info->fname = (char *) xmalloc (strlen(s)+1);
strcpy (info->fname, s);
logf (LOG_DEBUG, "creating tempfile %s", info->fname);
- info->fd = open (info->fname, O_RDWR|O_CREAT, 0666);
+ info->fd = open (info->fname, O_BINARY|O_RDWR|O_CREAT, 0666);
if (info->fd == -1)
{
logf (LOG_FATAL|LOG_ERRNO, "open %s", info->fname);
exit (1);
}
+#endif
}
if (info->fname && info->fd != -1 && info->dirty)
{
- size_t r, count;
+ size_t count;
+ int r;
if (lseek (info->fd, info->pos_buf, SEEK_SET) == -1)
{
count = info->buf_size;
if (count > info->pos_end - info->pos_buf)
count = info->pos_end - info->pos_buf;
- if ((r = write (info->fd, info->buf_mem, count)) < count)
+ if ((r = write (info->fd, info->buf_mem, count)) < (int) count)
{
if (r == -1)
logf (LOG_FATAL|LOG_ERRNO, "read %s", info->fname);
close (info->fd);
info->fd = -1;
}
+ xfree (((struct rset_temp_rfd *)rfd)->buf);
+ xfree (rfd);
}
-static void r_delete (struct rset_control *ct)
+static void r_delete (RSET ct)
{
- struct rset_temp_info *info = ct->buf;
+ struct rset_temp_info *info = (struct rset_temp_info*) ct->buf;
if (info->fname)
unlink (info->fname);
- free (info->buf_mem);
- free (info->fname);
- free (info);
+ xfree (info->buf_mem);
+ logf (LOG_DEBUG, "r_delete: set size %ld", (long) info->pos_end);
+ if (info->fname)
+ {
+ logf (LOG_DEBUG, "r_delete: unlink %s", info->fname);
+ unlink (info->fname);
+ xfree (info->fname);
+ }
+ if (info->temp_path)
+ xfree (info->temp_path);
+ rset_term_destroy (ct->rset_terms[0]);
+ xfree (ct->rset_terms);
+ xfree (info);
}
/* r_reread:
if (info->fname)
{
- size_t r, count;
+ size_t count;
+ int r;
info->pos_border = info->pos_cur + info->buf_size;
if (info->pos_border > info->pos_end)
logf (LOG_FATAL|LOG_ERRNO, "lseek %s", info->fname);
exit (1);
}
- if ((r = read (info->fd, info->buf_mem, count)) < count)
+ if ((r = read (info->fd, info->buf_mem, count)) < (int) count)
{
if (r == -1)
logf (LOG_FATAL|LOG_ERRNO, "read %s", info->fname);
r_reread (rfd);
}
-static int r_count (struct rset_control *ct)
+static int r_count (RSET ct)
{
- struct rset_temp_info *info = ct->buf;
+ struct rset_temp_info *info = (struct rset_temp_info *) ct->buf;
return info->pos_end / info->key_size;
}
-static int r_read (RSFD rfd, void *buf)
+static int r_read (RSFD rfd, void *buf, int *term_index)
{
- struct rset_temp_info *info = ((struct rset_temp_rfd*)rfd)->info;
+ struct rset_temp_rfd *mrfd = (struct rset_temp_rfd*) rfd;
+ struct rset_temp_info *info = mrfd->info;
size_t nc = info->pos_cur + info->key_size;
memcpy (buf, info->buf_mem + (info->pos_cur - info->pos_buf),
info->key_size);
info->pos_cur = nc;
+ *term_index = 0;
+
+ if (*mrfd->countp == 0 || (*info->cmp)(buf, mrfd->buf) > 1)
+ {
+ memcpy (mrfd->buf, buf, mrfd->info->key_size);
+ (*mrfd->countp)++;
+ }
return 1;
}
{
r_flush (rfd, 1);
info->pos_buf = info->pos_cur;
- r_reread (rfd);
- info->dirty = 1;
+ if (info->pos_buf < info->pos_end)
+ r_reread (rfd);
}
+ info->dirty = 1;
memcpy (info->buf_mem + (info->pos_cur - info->pos_buf), buf,
info->key_size);
info->pos_cur = nc;
info->pos_border = info->pos_end = nc;
return 1;
}
-
projects / idzebra-moved-to-github.git / blobdiff