2 * Copyright (c) 1990, 1993, 1994
3 * The Regents of the University of California. All rights reserved.
5 * This code is derived from software contributed to Berkeley by
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. All advertising materials mentioning features or use of this software
17 * must display the following acknowledgement:
18 * This product includes software developed by the University of
19 * California, Berkeley and its contributors.
20 * 4. Neither the name of the University nor the names of its contributors
21 * may be used to endorse or promote products derived from this software
22 * without specific prior written permission.
24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36 * @(#)bt_seq.c 8.7 (Berkeley) 7/20/94
39 #include <sys/types.h>
49 static int __bt_first __P((BTREE *, const DBT *, EPG *, int *));
50 static int __bt_seqadv __P((BTREE *, EPG *, int));
51 static int __bt_seqset __P((BTREE *, EPG *, DBT *, int));
54 * Sequential scan support.
56 * The tree can be scanned sequentially, starting from either end of the
57 * tree or from any specific key. A scan request before any scanning is
58 * done is initialized as starting from the least node.
63 * Btree sequential scan interface.
66 * dbp: pointer to access method
67 * key: key for positioning and return value
68 * data: data return value
69 * flags: R_CURSOR, R_FIRST, R_LAST, R_NEXT, R_PREV.
72 * RET_ERROR, RET_SUCCESS or RET_SPECIAL if there's no next key.
75 __bt_seq(dbp, key, data, flags)
86 /* Toss any page pinned across calls. */
87 if (t->bt_pinned != NULL) {
88 mpool_put(t->bt_mp, t->bt_pinned, 0);
93 * If scan unitialized as yet, or starting at a specific record, set
94 * the scan to a specific key. Both __bt_seqset and __bt_seqadv pin
95 * the page the cursor references if they're successful.
100 if (F_ISSET(&t->bt_cursor, CURS_INIT)) {
101 status = __bt_seqadv(t, &e, flags);
108 status = __bt_seqset(t, &e, key, flags);
115 if (status == RET_SUCCESS) {
116 __bt_setcur(t, e.page->pgno, e.index);
119 __bt_ret(t, &e, key, &t->bt_rkey, data, &t->bt_rdata, 0);
122 * If the user is doing concurrent access, we copied the
123 * key/data, toss the page.
125 if (F_ISSET(t, B_DB_LOCK))
126 mpool_put(t->bt_mp, e.page, 0);
128 t->bt_pinned = e.page;
135 * Set the sequential scan to a specific key.
139 * ep: storage for returned key
140 * key: key for initial scan position
141 * flags: R_CURSOR, R_FIRST, R_LAST, R_NEXT, R_PREV
144 * Pins the page the cursor references.
147 * RET_ERROR, RET_SUCCESS or RET_SPECIAL if there's no next key.
150 __bt_seqset(t, ep, key, flags)
161 * Find the first, last or specific key in the tree and point the
162 * cursor at it. The cursor may not be moved until a new key has
166 case R_CURSOR: /* Keyed scan. */
168 * Find the first instance of the key or the smallest key
169 * which is greater than or equal to the specified key.
171 if (key->data == NULL || key->size == 0) {
175 return (__bt_first(t, key, ep, &exact));
176 case R_FIRST: /* First record. */
178 /* Walk down the left-hand side of the tree. */
179 for (pg = P_ROOT;;) {
180 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL)
183 /* Check for an empty tree. */
184 if (NEXTINDEX(h) == 0) {
185 mpool_put(t->bt_mp, h, 0);
186 return (RET_SPECIAL);
189 if (h->flags & (P_BLEAF | P_RLEAF))
191 pg = GETBINTERNAL(h, 0)->pgno;
192 mpool_put(t->bt_mp, h, 0);
197 case R_LAST: /* Last record. */
199 /* Walk down the right-hand side of the tree. */
200 for (pg = P_ROOT;;) {
201 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL)
204 /* Check for an empty tree. */
205 if (NEXTINDEX(h) == 0) {
206 mpool_put(t->bt_mp, h, 0);
207 return (RET_SPECIAL);
210 if (h->flags & (P_BLEAF | P_RLEAF))
212 pg = GETBINTERNAL(h, NEXTINDEX(h) - 1)->pgno;
213 mpool_put(t->bt_mp, h, 0);
217 ep->index = NEXTINDEX(h) - 1;
220 return (RET_SUCCESS);
225 * Advance the sequential scan.
229 * flags: R_NEXT, R_PREV
232 * Pins the page the new key/data record is on.
235 * RET_ERROR, RET_SUCCESS or RET_SPECIAL if there's no next key.
238 __bt_seqadv(t, ep, flags)
250 * There are a couple of states that we can be in. The cursor has
251 * been initialized by the time we get here, but that's all we know.
256 * The cursor was deleted where there weren't any duplicate records,
257 * so the key was saved. Find out where that key would go in the
258 * current tree. It doesn't matter if the returned key is an exact
259 * match or not -- if it's an exact match, the record was added after
260 * the delete so we can just return it. If not, as long as there's
261 * a record there, return it.
263 if (F_ISSET(c, CURS_ACQUIRE))
264 return (__bt_first(t, &c->key, ep, &exact));
266 /* Get the page referenced by the cursor. */
267 if ((h = mpool_get(t->bt_mp, c->pg.pgno, 0)) == NULL)
271 * Find the next/previous record in the tree and point the cursor at
272 * it. The cursor may not be moved until a new key has been found.
275 case R_NEXT: /* Next record. */
277 * The cursor was deleted in duplicate records, and moved
278 * forward to a record that has yet to be returned. Clear
279 * that flag, and return the record.
281 if (F_ISSET(c, CURS_AFTER))
284 if (++index == NEXTINDEX(h)) {
286 mpool_put(t->bt_mp, h, 0);
288 return (RET_SPECIAL);
289 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL)
294 case R_PREV: /* Previous record. */
296 * The cursor was deleted in duplicate records, and moved
297 * backward to a record that has yet to be returned. Clear
298 * that flag, and return the record.
300 if (F_ISSET(c, CURS_BEFORE)) {
301 usecurrent: F_CLR(c, CURS_AFTER | CURS_BEFORE);
303 ep->index = c->pg.index;
304 return (RET_SUCCESS);
309 mpool_put(t->bt_mp, h, 0);
311 return (RET_SPECIAL);
312 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL)
314 index = NEXTINDEX(h) - 1;
322 return (RET_SUCCESS);
327 * Find the first entry.
333 * exactp: pointer to exact match flag
336 * The first entry in the tree greater than or equal to key,
337 * or RET_SPECIAL if no such key exists.
340 __bt_first(t, key, erval, exactp)
351 * Find any matching record; __bt_search pins the page.
353 * If it's an exact match and duplicates are possible, walk backwards
354 * in the tree until we find the first one. Otherwise, make sure it's
355 * a valid key (__bt_search may return an index just past the end of a
356 * page) and return it.
358 if ((ep = __bt_search(t, key, exactp)) == NULL)
361 if (F_ISSET(t, B_NODUPS)) {
363 return (RET_SUCCESS);
367 * Walk backwards, as long as the entry matches and there are
368 * keys left in the tree. Save a copy of each match in case
374 if (save.page->pgno != ep->page->pgno) {
375 mpool_put(t->bt_mp, save.page, 0);
378 save.index = ep->index;
381 * Don't unpin the page the last (or original) match
382 * was on, but make sure it's unpinned if an error
385 if (ep->index == 0) {
386 if (h->prevpg == P_INVALID)
388 if (h->pgno != save.page->pgno)
389 mpool_put(t->bt_mp, h, 0);
390 if ((h = mpool_get(t->bt_mp,
391 h->prevpg, 0)) == NULL) {
392 if (h->pgno == save.page->pgno)
398 ep->index = NEXTINDEX(h);
401 } while (__bt_cmp(t, key, ep) == 0);
404 * Reach here with the last page that was looked at pinned,
405 * which may or may not be the same as the last (or original)
406 * match page. If it's not useful, release it.
408 if (h->pgno != save.page->pgno)
409 mpool_put(t->bt_mp, h, 0);
412 return (RET_SUCCESS);
415 /* If at the end of a page, find the next entry. */
416 if (ep->index == NEXTINDEX(ep->page)) {
419 mpool_put(t->bt_mp, h, 0);
421 return (RET_SPECIAL);
422 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL)
428 return (RET_SUCCESS);
433 * Set the cursor to an entry in the tree.
441 __bt_setcur(t, pgno, index)
446 /* Lose any already deleted key. */
447 if (t->bt_cursor.key.data != NULL) {
448 free(t->bt_cursor.key.data);
449 t->bt_cursor.key.size = 0;
450 t->bt_cursor.key.data = NULL;
452 F_CLR(&t->bt_cursor, CURS_ACQUIRE | CURS_AFTER | CURS_BEFORE);
454 /* Update the cursor. */
455 t->bt_cursor.pg.pgno = pgno;
456 t->bt_cursor.pg.index = index;
457 F_SET(&t->bt_cursor, CURS_INIT);