1 /* $Id: rset.h,v 1.54 2005-06-06 21:31:08 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;
36 typedef struct rset *RSET;
40 struct ord_list *next;
44 * rset_term is all we need to know of a term to do ranking etc.
45 * As far as the rsets are concerned, it is just a dummy pointer to
49 char *name; /** the term itself in internal encoding (UTF-8/raw) */
50 char *flags; /** flags for rank method */
51 int type; /** Term_type from RPN Query. Actually this
52 is Z_Term_general, Z_Term_numeric,
53 Z_Term_characterString, ..
54 This info is used to return encoded term back for
57 RSET rset; /** the rset corresponding to this term */
58 void *rankpriv;/** private stuff for the ranking algorithm */
62 typedef struct rset_term *TERMID;
63 TERMID rset_term_create (const char *name, int length, const char *flags,
64 int type, NMEM nmem, struct ord_list *ol);
66 /** rsfd is a "file descriptor" for reading from a rset */
67 struct rsfd { /* the stuff common to all rsfd's. */
68 RSET rset; /* ptr to the rset this FD is opened to */
69 void *priv; /* private parameters for this type */
70 RSFD next; /* to keep lists of used/free rsfd's */
77 * rset_control has function pointers to all the important functions
78 * of a rset. Each type of rset will have its own control block, pointing
79 * to the functions for that type. They all have their own create function
80 * which is not part of the control block, as it takes different args for
85 /** text description of set type (for debugging) */
87 /* RSET rs_something_create(const struct rset_control *sel, ...); */
88 void (*f_delete)(RSET ct);
90 /** recursively fills the terms array with terms. call with curterm=0 */
91 /* always counts them all into cur, but of course won't touch the term */
92 /* array past max. You can use this to count, set max=0 */
93 void (*f_getterms)(RSET ct, TERMID *terms, int maxterms, int *curterm);
95 RSFD (*f_open)(RSET ct, int wflag);
96 void (*f_close)(RSFD rfd);
97 /** forward behaves like a read, but it skips some items first */
98 int (*f_forward)(RSFD rfd, void *buf, TERMID *term, const void *untilbuf);
99 void (*f_pos)(RSFD rfd, double *current, double *total);
100 /* returns -1,-1 if pos function not implemented for this type */
101 int (*f_read)(RSFD rfd, void *buf, TERMID *term);
102 int (*f_write)(RSFD rfd, const void *buf);
105 /** rset_default_forward implements a generic forward with a read-loop */
106 int rset_default_forward(RSFD rfd, void *buf, TERMID *term,
107 const void *untilbuf);
109 /** rset_default_read implements a generic read */
110 int rset_default_read(RSFD rfd, void *buf, TERMID *term);
112 void rset_get_one_term(RSET ct,TERMID *terms,int maxterms,int *curterm);
115 * key_control contains all there is to know about the keys stored in
116 * an isam, and therefore operated by the rsets. Other than this info,
117 * all we assume is that all keys are the same size, and they can be
120 struct rset_key_control {
123 int scope; /* default for what level we operate (book/chapter/verse) on*/
124 /* usual sysno/seqno is 2 */
125 int (*cmp)(const void *p1, const void *p2);
126 void (*key_logdump_txt) (int logmask, const void *p, const char *txt);
127 zint (*getseq)(const void *p);
128 int (*filter_func)(const void *p, void *data);
130 void (*inc)(struct rset_key_control *kc);
131 void (*dec)(struct rset_key_control *kc);
132 /* FIXME - Should not need a getseq, it won't make much sense with */
133 /* higher-order keys. Use a (generalized) cmp instead, or something */
134 /* FIXME - decode and encode, and lots of other stuff */
138 * A rset is an ordered sequence of keys, either directly from an underlaying
139 * isam, or from one of the higher-level operator rsets (and, or, ...).
140 * Actually, it is "virtual base class", no pure rsets exist in the system,
141 * they all are of some derived type.
145 const struct rset_control *control;
146 struct rset_key_control *keycontrol;
147 int refcount; /* reference count */
148 void *priv; /* stuff private to the given type of rset */
149 NMEM nmem; /* nibble memory for various allocs */
150 RSFD free_list; /* all rfd's allocated but not currently in use */
151 RSFD use_list; /* all rfd's in use */
152 int scope; /* On what level do we count hits and compare them? */
153 TERMID term; /* the term thing for ranking etc */
161 /* rset is a "virtual base class", which will never exist on its own
162 * all instances are rsets of some specific type, like rsisamb, or rsbool
163 * They keep their own stuff behind the priv pointer. */
165 /* On the old sysno-seqno type isams, the scope was hard-coded to be 2.
166 * This means that we count hits on the sysno level, and when matching an
167 * 'and', we consider it a match if both term occur within the same sysno.
168 * In more complex isams we can specify on what level we wish to do the
169 * matching and counting of hits. For example, we can have book / chapter /
170 * verse, and a seqno. Scope 2 means then "give me all verses that match",
171 * 3 would be chapters, 4 books.
172 * The resolution tells how much of the occurences we need to return. If we
173 * are doing some sort of proximity, we need to get the seqnos of all
174 * occurences, whereas if we are only counting hits, we do not need anything
175 * below the scope. Again 1 is seqnos, 2 sysnos (or verses), 3 books, etc.
178 RSFD rfd_create_base(RSET rs);
179 int rfd_is_last(RSFD rfd);
181 RSET rset_create_base(const struct rset_control *sel,
183 struct rset_key_control *kcontrol,
186 int no_children, RSET *children);
188 void rset_delete(RSET rs);
189 RSET rset_dup (RSET rs);
190 void rset_close(RSFD rfd);
193 #define RSETF_WRITE 1
194 /* RSFD rset_open(RSET rs, int wflag); */
195 #define rset_open(rs, wflag) (*(rs)->control->f_open)((rs), (wflag))
197 /* int rset_forward(RSFD rfd, void *buf, TERMID term, void *untilbuf); */
198 #define rset_forward(rfd, buf, term, untilbuf) \
199 rset_default_forward((rfd), (buf), (term), (untilbuf))
201 /* void rset_getterms(RSET ct, TERMID *terms, int maxterms, int *curterm); */
202 #define rset_getterms(ct, terms, maxterms, curterm) \
203 (*(ct)->control->f_getterms)((ct),(terms),(maxterms),(curterm))
205 /* int rset_pos(RSFD fd, double *current, double *total); */
206 #define rset_pos(rfd,cur,tot) \
207 (*(rfd)->rset->control->f_pos)((rfd),(cur),(tot))
209 /* int rset_read(RSFD rfd, void *buf, TERMID term); */
210 #define rset_read(rfd, buf, term) rset_default_read((rfd), (buf), (term))
212 /* int rset_write(RSFD rfd, const void *buf); */
213 #define rset_write(rfd, buf) (*(rfd)->rset->control->f_write)((rfd), (buf))
215 /* int rset_type (RSET) */
216 #define rset_type(rs) ((rs)->control->desc)
218 /** rset_count counts or estimates the keys in it*/
219 zint rset_count(RSET rs);
221 RSET rstemp_create(NMEM nmem, struct rset_key_control *kcontrol,
222 int scope, const char *temp_path, TERMID term);
224 RSET rsnull_create(NMEM nmem, struct rset_key_control *kcontrol, TERMID term);
226 RSET rsbool_create_and(NMEM nmem, struct rset_key_control *kcontrol,
227 int scope, RSET rset_l, RSET rset_r);
229 RSET rsbool_create_or(NMEM nmem, struct rset_key_control *kcontrol,
230 int scope, RSET rset_l, RSET rset_r);
232 RSET rsbool_create_not(NMEM nmem, struct rset_key_control *kcontrol,
233 int scope, RSET rset_l, RSET rset_r);
235 RSET rsbetween_create(NMEM nmem, struct rset_key_control *kcontrol,
236 int scope, RSET rset_l, RSET rset_m, RSET rset_r,
239 RSET rsmulti_or_create(NMEM nmem, struct rset_key_control *kcontrol,
240 int scope, TERMID termid, int no_rsets, RSET* rsets);
242 RSET rsmulti_and_create(NMEM nmem, struct rset_key_control *kcontrol,
243 int scope, int no_rsets, RSET* rsets);
245 RSET rsprox_create(NMEM nmem, struct rset_key_control *kcontrol,
246 int scope, int rset_no, RSET *rset,
247 int ordered, int exclusion, int relation, int distance);
249 RSET rsisamb_create(NMEM nmem, struct rset_key_control *kcontrol,
250 int scope, ISAMB is, ISAM_P pos, TERMID term);
252 RSET rsisamc_create(NMEM nmem, struct rset_key_control *kcontrol,
253 int scope, ISAMC is, ISAM_P pos, TERMID term);
255 RSET rsisams_create(NMEM nmem, struct rset_key_control *kcontrol,
256 int scope, ISAMS is, ISAM_P pos, TERMID term);
258 void rset_visit(RSET rset, int level);