X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=rset%2Frstemp.c;h=21b0e79e0d93b78a1c5117309c2e6b7da50151d0;hb=74eec9c27962c38e3a6f0983f78adf7e9d7de0eb;hp=34dd15251dd84425daa30ae29bac1e5aa4b5e4ab;hpb=c427496a1254a039887e02c8bb7ce1d673bb251b;p=idzebra-moved-to-github.git diff --git a/rset/rstemp.c b/rset/rstemp.c index 34dd152..21b0e79 100644 --- a/rset/rstemp.c +++ b/rset/rstemp.c @@ -1,10 +1,76 @@ /* - * Copyright (C) 1994, Index Data I/S + * Copyright (C) 1994-1997, Index Data I/S * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * * $Log: rstemp.c,v $ - * Revision 1.3 1995-09-04 15:20:40 adam + * Revision 1.22 1997-10-31 12:38:12 adam + * Bug fix: added missing xfree() call. + * + * Revision 1.21 1997/09/17 12:19:23 adam + * Zebra version corresponds to YAZ version 1.4. + * Changed Zebra server so that it doesn't depend on global common_resource. + * + * Revision 1.20 1997/09/09 13:38:17 adam + * Partial port to WIN95/NT. + * + * Revision 1.19 1997/09/04 13:58:57 adam + * Added O_BINARY for open calls. + * + * Revision 1.18 1996/10/29 13:54:52 adam + * Changed name of setting tempSetDir to setTmpDir. + * + * Revision 1.17 1995/12/11 09:15:28 adam + * New set types: sand/sor/snot - ranked versions of and/or/not in + * ranked/semi-ranked result sets. + * Note: the snot not finished yet. + * New rset member: flag. + * Bug fix: r_delete in rsrel.c did free bad memory block. + * + * Revision 1.16 1995/11/28 14:47:02 adam + * New setting: tempSetPath. Location of temporary result sets. + * + * Revision 1.15 1995/10/12 12:41:58 adam + * Private info (buf) moved from struct rset_control to struct rset. + * Bug fixes in relevance. + * + * Revision 1.14 1995/10/10 14:00:04 adam + * Function rset_open changed its wflag parameter to general flags. + * + * Revision 1.13 1995/10/06 14:38:06 adam + * New result set method: r_score. + * Local no (sysno) and score is transferred to retrieveCtrl. + * + * Revision 1.12 1995/09/28 09:52:11 adam + * xfree/xmalloc used everywhere. + * + * Revision 1.11 1995/09/18 14:17:56 adam + * Bug fixes. + * + * 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 @@ -15,25 +81,34 @@ * */ +#include +#include +#ifdef WINDOWS +#include +#else +#include +#endif +#include +#include #include -#include +#include #include -static struct rset_control *r_create(const struct rset_control *sel, - void *parms); -static int r_open(struct rset_control *ct, int wflag); -static void r_close(struct rset_control *ct); -static void r_delete(struct rset_control *ct); -static void r_rewind(struct rset_control *ct); -static int r_count(struct rset_control *ct); -static int r_read(); -static int r_write(); +static void *r_create(const struct rset_control *sel, void *parms, + int *flags); +static RSFD r_open (RSET ct, int flag); +static void r_close (RSFD rfd); +static void r_delete (RSET ct); +static void r_rewind (RSFD rfd); +static int r_count (RSET ct); +static int r_read (RSFD rfd, void *buf); +static int r_write (RSFD rfd, const void *buf); +static int r_score (RSFD rfd, int *score); static const rset_control control = { - "Temporary set", - 0, + "temp", r_create, r_open, r_close, @@ -41,68 +116,259 @@ static const rset_control control = r_rewind, r_count, r_read, - r_write + r_write, + r_score }; const rset_control *rset_kind_temp = &control; -struct rset_temp_private { +struct rset_temp_info { int fd; char *fname; - size_t key_size; - char *buf_mem; - size_t buf_size; - size_t pos_end; - size_t pos_cur; - size_t pos_buf; + size_t key_size; /* key size */ + char *buf_mem; /* window buffer */ + size_t buf_size; /* size of window */ + size_t pos_end; /* last position in set */ + size_t pos_cur; /* current position in set */ + 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 */ + char *temp_path; +}; + +struct rset_temp_rfd { + struct rset_temp_info *info; + struct rset_temp_rfd *next; }; -static struct rset_control *r_create(const struct rset_control *sel, - void *parms) +static void *r_create(const struct rset_control *sel, void *parms, int *flags) { - rset_control *newct; rset_temp_parms *temp_parms = parms; - struct rset_temp_private *info; - - logf (LOG_DEBUG, "ritemp_create(%s)", sel->desc); - newct = xmalloc(sizeof(*newct)); - memcpy(newct, sel, sizeof(*sel)); - newct->buf = xmalloc (sizeof(struct rset_temp_private)); - info = newct->buf; - + struct rset_temp_info *info; + + 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_size = 4096; info->buf_mem = xmalloc (info->buf_size); info->pos_cur = 0; info->pos_end = 0; info->pos_buf = 0; + info->dirty = 0; + if (!temp_parms->temp_path) + info->temp_path = NULL; + else + { + info->temp_path = xmalloc (strlen(temp_parms->temp_path)+1); + strcpy (info->temp_path, temp_parms->temp_path); + } + + return info; +} + +static RSFD r_open (RSET ct, int flag) +{ + struct rset_temp_info *info = ct->buf; + struct rset_temp_rfd *rfd; + + assert (info->fd == -1); + if (info->fname) + { + if (flag & RSETF_WRITE) + info->fd = open (info->fname, O_BINARY|O_RDWR|O_CREAT, 0666); + else + 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->info = info; + r_rewind (rfd); + return rfd; +} + +/* r_flush: + flush current window to file if file is assocated with set + */ +static void r_flush (RSFD rfd, int mk) +{ + struct rset_temp_info *info = ((struct rset_temp_rfd*) rfd)->info; + + if (!info->fname && mk) + { + char *s = (char*) tempnam (info->temp_path, "zrs"); + + info->fname = xmalloc (strlen(s)+1); + strcpy (info->fname, s); + + 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); + } + } + if (info->fname && info->fd != -1 && info->dirty) + { + size_t r, count; + + if (lseek (info->fd, info->pos_buf, SEEK_SET) == -1) + { + logf (LOG_FATAL|LOG_ERRNO, "lseek %s", info->fname); + exit (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 == -1) + logf (LOG_FATAL|LOG_ERRNO, "read %s", info->fname); + else + logf (LOG_FATAL, "write of %ld but got %ld", + (long) count, (long) r); + exit (1); + } + info->dirty = 0; + } +} + +static void r_close (RSFD rfd) +{ + struct rset_temp_info *info = ((struct rset_temp_rfd*)rfd)->info; + + r_flush (rfd, 0); + if (info->fname && info->fd != -1) + { + close (info->fd); + info->fd = -1; + } + xfree (rfd); +} + +static void r_delete (RSET ct) +{ + struct rset_temp_info *info = ct->buf; + + if (info->fname) + unlink (info->fname); + 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); + xfree (info); +} + +/* r_reread: + read from file to window if file is assocated with set - + indicated by fname + */ +static void r_reread (RSFD rfd) +{ + struct rset_temp_info *info = ((struct rset_temp_rfd*)rfd)->info; - return newct; + if (info->fname) + { + size_t r, count; + + 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 (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)) < count) + { + if (r == -1) + logf (LOG_FATAL|LOG_ERRNO, "read %s", info->fname); + else + logf (LOG_FATAL, "read of %ld but got %ld", + (long) count, (long) r); + exit (1); + } + } + } + else + info->pos_border = info->pos_end; } -static int r_open(struct rset_control *ct, int wflag) +static void r_rewind (RSFD rfd) { - struct rset_temp_private *info = ct->buf; + struct rset_temp_info *info = ((struct rset_temp_rfd*)rfd)->info; + + r_flush (rfd, 0); info->pos_cur = 0; info->pos_buf = 0; + r_reread (rfd); } -static void r_close(struct rset_control *ct) -{} +static int r_count (RSET ct) +{ + struct rset_temp_info *info = ct->buf; -static void r_delete(struct rset_control *ct) -{} + return info->pos_end / info->key_size; +} -static void r_rewind(struct rset_control *ct) -{} +static int r_read (RSFD rfd, void *buf) +{ + struct rset_temp_info *info = ((struct rset_temp_rfd*)rfd)->info; -static int r_count(struct rset_control *ct) -{} + size_t nc = info->pos_cur + info->key_size; -static int r_read() -{} + if (nc > info->pos_border) + { + if (nc > info->pos_end) + return 0; + r_flush (rfd, 0); + info->pos_buf = info->pos_cur; + r_reread (rfd); + } + memcpy (buf, info->buf_mem + (info->pos_cur - info->pos_buf), + info->key_size); + info->pos_cur = nc; + return 1; +} -static int r_write() -{} +static int r_write (RSFD rfd, const void *buf) +{ + struct rset_temp_info *info = ((struct rset_temp_rfd*)rfd)->info; + + size_t nc = info->pos_cur + info->key_size; + + if (nc > info->pos_buf + info->buf_size) + { + r_flush (rfd, 1); + info->pos_buf = info->pos_cur; + 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; + if (nc > info->pos_end) + info->pos_border = info->pos_end = nc; + return 1; +} + +static int r_score (RSFD rfd, int *score) +{ + *score = -1; + return -1; +}