1 /* $Id: rset.h,v 1.52 2005-05-24 11:35:41 adam Exp $
2 Copyright (C) 1995-2005
5 This file is part of the Zebra server.
7 Zebra is free software; you can redistribute it and/or modify it under
8 the terms of the GNU General Public License as published by the Free
9 Software Foundation; either version 2, or (at your option) any later
12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
14 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
17 You should have received a copy of the GNU General Public License
18 along with Zebra; see the file LICENSE.zebra. If not, write to the
19 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
26 #include <yaz/yaz-util.h>
27 /* unfortunately we need the isam includes here, for the arguments for */
29 #include <idzebra/isamb.h>
30 #include <idzebra/isamc.h>
31 #include <idzebra/isams.h>
35 typedef struct rsfd *RSFD; /* Rset "file descriptor" */
36 typedef struct rset *RSET; /* Result set */
40 * rset_term is all we need to know of a term to do ranking etc.
41 * As far as the rsets are concerned, it is just a dummy pointer to
46 /** the term itself */
50 /** the rset corresponding to this term */
52 /** private stuff for the ranking algorithm */
56 typedef struct rset_term *TERMID;
57 TERMID rset_term_create (const char *name, int length, const char *flags,
60 /** rsfd is a "file descriptor" for reading from a rset */
61 struct rsfd { /* the stuff common to all rsfd's. */
62 RSET rset; /* ptr to the rset this FD is opened to */
63 void *priv; /* private parameters for this type */
64 RSFD next; /* to keep lists of used/free rsfd's */
71 * rset_control has function pointers to all the important functions
72 * of a rset. Each type of rset will have its own control block, pointing
73 * to the functions for that type. They all have their own create function
74 * which is not part of the control block, as it takes different args for
79 /** text description of set type (for debugging) */
81 /* RSET rs_something_create(const struct rset_control *sel, ...); */
82 void (*f_delete)(RSET ct);
84 /** recursively fills the terms array with terms. call with curterm=0 */
85 /* always counts them all into cur, but of course won't touch the term */
86 /* array past max. You can use this to count, set max=0 */
87 void (*f_getterms)(RSET ct, TERMID *terms, int maxterms, int *curterm);
89 RSFD (*f_open)(RSET ct, int wflag);
90 void (*f_close)(RSFD rfd);
91 /** forward behaves like a read, but it skips some items first */
92 int (*f_forward)(RSFD rfd, void *buf, TERMID *term, const void *untilbuf);
93 void (*f_pos)(RSFD rfd, double *current, double *total);
94 /* returns -1,-1 if pos function not implemented for this type */
95 int (*f_read)(RSFD rfd, void *buf, TERMID *term);
96 int (*f_write)(RSFD rfd, const void *buf);
99 /** rset_default_forward implements a generic forward with a read-loop */
100 int rset_default_forward(RSFD rfd, void *buf, TERMID *term,
101 const void *untilbuf);
103 /** rset_default_read implements a generic read */
104 int rset_default_read(RSFD rfd, void *buf, TERMID *term);
106 /** rset_get_no_terms is a getterms function for those that don't have any */
107 void rset_get_no_terms(RSET ct, TERMID *terms, int maxterms, int *curterm);
110 * rset_get_one_term is a getterms function for those rsets that have
111 * exactly one term, like all rsisamX types.
113 void rset_get_one_term(RSET ct,TERMID *terms,int maxterms,int *curterm);
116 * key_control contains all there is to know about the keys stored in
117 * an isam, and therefore operated by the rsets. Other than this info,
118 * all we assume is that all keys are the same size, and they can be
121 struct rset_key_control {
124 int scope; /* default for what level we operate (book/chapter/verse) on*/
125 /* usual sysno/seqno is 2 */
126 int (*cmp)(const void *p1, const void *p2);
127 void (*key_logdump_txt) (int logmask, const void *p, const char *txt);
128 zint (*getseq)(const void *p);
129 int (*filter_func)(const void *p, void *data);
131 void (*inc)(struct rset_key_control *kc);
132 void (*dec)(struct rset_key_control *kc);
133 /* FIXME - Should not need a getseq, it won't make much sense with */
134 /* higher-order keys. Use a (generalized) cmp instead, or something */
135 /* FIXME - decode and encode, and lots of other stuff */
139 * A rset is an ordered sequence of keys, either directly from an underlaying
140 * isam, or from one of the higher-level operator rsets (and, or, ...).
141 * Actually, it is "virtual base class", no pure rsets exist in the system,
142 * they all are of some derived type.
146 const struct rset_control *control;
147 struct rset_key_control *keycontrol;
148 int refcount; /* reference count */
149 void *priv; /* stuff private to the given type of rset */
150 NMEM nmem; /* nibble memory for various allocs */
151 char my_nmem; /* Should the nmem be destroyed with the rset? */
152 /* 1 if created with it, 0 if passed from above */
153 RSFD free_list; /* all rfd's allocated but not currently in use */
154 RSFD use_list; /* all rfd's in use */
155 int scope; /* On what level do we count hits and compare them? */
156 TERMID term; /* the term thing for ranking etc */
164 /* rset is a "virtual base class", which will never exist on its own
165 * all instances are rsets of some specific type, like rsisamb, or rsbool
166 * They keep their own stuff behind the priv pointer. */
168 /* On the old sysno-seqno type isams, the scope was hard-coded to be 2.
169 * This means that we count hits on the sysno level, and when matching an
170 * 'and', we consider it a match if both term occur within the same sysno.
171 * In more complex isams we can specify on what level we wish to do the
172 * matching and counting of hits. For example, we can have book / chapter /
173 * verse, and a seqno. Scope 2 means then "give me all verses that match",
174 * 3 would be chapters, 4 books.
175 * The resolution tells how much of the occurences we need to return. If we
176 * are doing some sort of proximity, we need to get the seqnos of all
177 * occurences, whereas if we are only counting hits, we do not need anything
178 * below the scope. Again 1 is seqnos, 2 sysnos (or verses), 3 books, etc.
181 RSFD rfd_create_base(RSET rs);
182 int rfd_is_last(RSFD rfd);
184 RSET rset_create_base(const struct rset_control *sel,
186 struct rset_key_control *kcontrol,
189 int no_children, RSET *children);
191 void rset_delete(RSET rs);
192 RSET rset_dup (RSET rs);
193 void rset_close(RSFD rfd);
196 #define RSETF_WRITE 1
197 /* RSFD rset_open(RSET rs, int wflag); */
198 #define rset_open(rs, wflag) (*(rs)->control->f_open)((rs), (wflag))
200 /* int rset_forward(RSFD rfd, void *buf, TERMID term, void *untilbuf); */
201 #define rset_forward(rfd, buf, term, untilbuf) \
202 rset_default_forward((rfd), (buf), (term), (untilbuf))
204 /* void rset_getterms(RSET ct, TERMID *terms, int maxterms, int *curterm); */
205 #define rset_getterms(ct, terms, maxterms, curterm) \
206 (*(ct)->control->f_getterms)((ct),(terms),(maxterms),(curterm))
208 /* int rset_pos(RSFD fd, double *current, double *total); */
209 #define rset_pos(rfd,cur,tot) \
210 (*(rfd)->rset->control->f_pos)((rfd),(cur),(tot))
212 /* int rset_read(RSFD rfd, void *buf, TERMID term); */
213 #define rset_read(rfd, buf, term) rset_default_read((rfd), (buf), (term))
215 /* int rset_write(RSFD rfd, const void *buf); */
216 #define rset_write(rfd, buf) (*(rfd)->rset->control->f_write)((rfd), (buf))
218 /* int rset_type (RSET) */
219 #define rset_type(rs) ((rs)->control->desc)
221 /** rset_count counts or estimates the keys in it*/
222 zint rset_count(RSET rs);
224 RSET rstemp_create(NMEM nmem, struct rset_key_control *kcontrol,
225 int scope, const char *temp_path, TERMID term);
227 RSET rsnull_create(NMEM nmem, struct rset_key_control *kcontrol, TERMID term);
229 RSET rsbool_create_and(NMEM nmem, struct rset_key_control *kcontrol,
230 int scope, RSET rset_l, RSET rset_r);
232 RSET rsbool_create_or(NMEM nmem, struct rset_key_control *kcontrol,
233 int scope, RSET rset_l, RSET rset_r);
235 RSET rsbool_create_not(NMEM nmem, struct rset_key_control *kcontrol,
236 int scope, RSET rset_l, RSET rset_r);
238 RSET rsbetween_create(NMEM nmem, struct rset_key_control *kcontrol,
239 int scope, RSET rset_l, RSET rset_m, RSET rset_r,
242 RSET rsmulti_or_create(NMEM nmem, struct rset_key_control *kcontrol,
243 int scope, TERMID termid, int no_rsets, RSET* rsets);
245 RSET rsmulti_and_create(NMEM nmem, struct rset_key_control *kcontrol,
246 int scope, int no_rsets, RSET* rsets);
248 RSET rsprox_create(NMEM nmem, struct rset_key_control *kcontrol,
249 int scope, int rset_no, RSET *rset,
250 int ordered, int exclusion, int relation, int distance);
252 RSET rsisamb_create(NMEM nmem, struct rset_key_control *kcontrol,
253 int scope, ISAMB is, ISAM_P pos, TERMID term);
255 RSET rsisamc_create(NMEM nmem, struct rset_key_control *kcontrol,
256 int scope, ISAMC is, ISAM_P pos, TERMID term);
258 RSET rsisams_create(NMEM nmem, struct rset_key_control *kcontrol,
259 int scope, ISAMS is, ISAM_P pos, TERMID term);
261 void rset_visit(RSET rset, int level);