X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=rset%2Frstemp.c;h=bd4f83bb7ede1af399774846a227273eb69273ed;hb=cc4fcee50354f71ed32a586de6d226b61139819e;hp=6a92d8371a1bb9057175a34cb881817242f654aa;hpb=7c4f8bf9516beb806a1e51467d2e10f86a4c21b7;p=idzebra-moved-to-github.git diff --git a/rset/rstemp.c b/rset/rstemp.c index 6a92d83..bd4f83b 100644 --- a/rset/rstemp.c +++ b/rset/rstemp.c @@ -1,62 +1,37 @@ /* - * 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 1.9 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 1.29 2002-03-15 20:11:36 adam Exp $ */ #include #include +#ifdef WIN32 +#include +#else #include +#endif +#include #include #include -#include +#include #include -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, @@ -64,10 +39,10 @@ static const rset_control control = 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; @@ -80,6 +55,8 @@ struct rset_temp_info { 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; }; struct rset_temp_rfd { @@ -87,52 +64,56 @@ struct rset_temp_rfd { struct rset_temp_rfd *next; }; -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 = 1024; - 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; - - return newct; + info->hits = -1; + 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 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 (ct); + r_rewind (rfd); return rfd; } @@ -145,21 +126,38 @@ static void r_flush (RSFD rfd, int mk) if (!info->fname && mk) { - char *s = (char*) tempnam (NULL, "zrs"); +#if HAVE_MKSTEMP + char template[1024]; + + sprintf (template, "%s/zrsXXXXXX", info->temp_path); - info->fname = xmalloc (strlen(s)+1); + 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); - info->fd = open (info->fname, O_RDWR|O_CREAT, 0666); + logf (LOG_DEBUG, "creating tempfile %s", info->fname); + 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) { @@ -169,7 +167,7 @@ static void r_flush (RSFD rfd, int mk) 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); @@ -192,17 +190,28 @@ static void r_close (RSFD rfd) close (info->fd); info->fd = -1; } + 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: @@ -215,14 +224,21 @@ static void r_reread (RSFD rfd) 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) info->pos_border = info->pos_end; count = info->pos_border - info->pos_buf; if (count > 0) - if ((r = read (info->fd, info->buf_mem, count)) < count) + { + if (lseek (info->fd, info->pos_buf, SEEK_SET) == -1) + { + logf (LOG_FATAL|LOG_ERRNO, "lseek %s", info->fname); + exit (1); + } + if ((r = read (info->fd, info->buf_mem, count)) < (int) count) { if (r == -1) logf (LOG_FATAL|LOG_ERRNO, "read %s", info->fname); @@ -231,6 +247,7 @@ static void r_reread (RSFD rfd) (long) count, (long) r); exit (1); } + } } else info->pos_border = info->pos_end; @@ -246,14 +263,14 @@ static void r_rewind (RSFD rfd) 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; @@ -270,6 +287,7 @@ static int r_read (RSFD rfd, void *buf) memcpy (buf, info->buf_mem + (info->pos_cur - info->pos_buf), info->key_size); info->pos_cur = nc; + *term_index = 0; return 1; } @@ -283,9 +301,10 @@ static int r_write (RSFD rfd, const void *buf) { 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; @@ -293,4 +312,3 @@ static int r_write (RSFD rfd, const void *buf) info->pos_border = info->pos_end = nc; return 1; } -