4 ** The author disclaims copyright to this source code. In place of
5 ** a legal notice, here is a blessing:
7 ** May you do good and not evil.
8 ** May you find forgiveness for yourself and forgive others.
9 ** May you share freely, never taking more than you give.
11 *************************************************************************
12 ** This file contains code used to implement the VACUUM command.
14 ** Most of the code in this file may be omitted by defining the
15 ** SQLITE_OMIT_VACUUM macro.
17 #include "sqliteInt.h"
20 #if !defined(SQLITE_OMIT_VACUUM) && !defined(SQLITE_OMIT_ATTACH)
22 ** Finalize a prepared statement. If there was an error, store the
23 ** text of the error message in *pzErrMsg. Return the result code.
25 static int vacuumFinalize(sqlite3
*db
, sqlite3_stmt
*pStmt
, char **pzErrMsg
){
27 rc
= sqlite3VdbeFinalize((Vdbe
*)pStmt
);
29 sqlite3SetString(pzErrMsg
, db
, sqlite3_errmsg(db
));
35 ** Execute zSql on database db. Return an error code.
37 static int execSql(sqlite3
*db
, char **pzErrMsg
, const char *zSql
){
43 if( SQLITE_OK
!=sqlite3_prepare(db
, zSql
, -1, &pStmt
, 0) ){
44 sqlite3SetString(pzErrMsg
, db
, sqlite3_errmsg(db
));
45 return sqlite3_errcode(db
);
47 VVA_ONLY( rc
= ) sqlite3_step(pStmt
);
48 assert( rc
!=SQLITE_ROW
|| (db
->flags
&SQLITE_CountRows
) );
49 return vacuumFinalize(db
, pStmt
, pzErrMsg
);
53 ** Execute zSql on database db. The statement returns exactly
54 ** one column. Execute this as SQL on the same database.
56 static int execExecSql(sqlite3
*db
, char **pzErrMsg
, const char *zSql
){
60 rc
= sqlite3_prepare(db
, zSql
, -1, &pStmt
, 0);
61 if( rc
!=SQLITE_OK
) return rc
;
63 while( SQLITE_ROW
==sqlite3_step(pStmt
) ){
64 rc
= execSql(db
, pzErrMsg
, (char*)sqlite3_column_text(pStmt
, 0));
66 vacuumFinalize(db
, pStmt
, pzErrMsg
);
71 return vacuumFinalize(db
, pStmt
, pzErrMsg
);
75 ** The VACUUM command is used to clean up the database,
76 ** collapse free space, etc. It is modelled after the VACUUM command
77 ** in PostgreSQL. The VACUUM command works as follows:
79 ** (1) Create a new transient database file
80 ** (2) Copy all content from the database being vacuumed into
81 ** the new transient database file
82 ** (3) Copy content from the transient database back into the
85 ** The transient database requires temporary disk space approximately
86 ** equal to the size of the original database. The copy operation of
87 ** step (3) requires additional temporary disk space approximately equal
88 ** to the size of the original database for the rollback journal.
89 ** Hence, temporary disk space that is approximately 2x the size of the
90 ** original database is required. Every page of the database is written
91 ** approximately 3 times: Once for step (2) and twice for step (3).
92 ** Two writes per page are required in step (3) because the original
93 ** database content must be written into the rollback journal prior to
94 ** overwriting the database with the vacuumed content.
96 ** Only 1x temporary space and only 1x writes would be required if
97 ** the copy of step (3) were replace by deleting the original database
98 ** and renaming the transient database as the original. But that will
99 ** not work if other processes are attached to the original database.
100 ** And a power loss in between deleting the original and renaming the
101 ** transient would cause the database file to appear to be deleted
104 void sqlite3Vacuum(Parse
*pParse
){
105 Vdbe
*v
= sqlite3GetVdbe(pParse
);
107 sqlite3VdbeAddOp2(v
, OP_Vacuum
, 0, 0);
108 sqlite3VdbeUsesBtree(v
, 0);
114 ** This routine implements the OP_Vacuum opcode of the VDBE.
116 int sqlite3RunVacuum(char **pzErrMsg
, sqlite3
*db
){
117 int rc
= SQLITE_OK
; /* Return code from service routines */
118 Btree
*pMain
; /* The database being vacuumed */
119 Btree
*pTemp
; /* The temporary database we vacuum into */
120 char *zSql
= 0; /* SQL statements */
121 int saved_flags
; /* Saved value of the db->flags */
122 int saved_nChange
; /* Saved value of db->nChange */
123 int saved_nTotalChange
; /* Saved value of db->nTotalChange */
124 void (*saved_xTrace
)(void*,const char*); /* Saved db->xTrace */
125 Db
*pDb
= 0; /* Database to detach at end of vacuum */
126 int isMemDb
; /* True if vacuuming a :memory: database */
127 int nRes
; /* Bytes of reserved space at the end of each page */
128 int nDb
; /* Number of attached databases */
130 if( !db
->autoCommit
){
131 sqlite3SetString(pzErrMsg
, db
, "cannot VACUUM from within a transaction");
134 if( db
->nVdbeActive
>1 ){
135 sqlite3SetString(pzErrMsg
, db
,"cannot VACUUM - SQL statements in progress");
139 /* Save the current value of the database flags so that it can be
140 ** restored before returning. Then set the writable-schema flag, and
141 ** disable CHECK and foreign key constraints. */
142 saved_flags
= db
->flags
;
143 saved_nChange
= db
->nChange
;
144 saved_nTotalChange
= db
->nTotalChange
;
145 saved_xTrace
= db
->xTrace
;
146 db
->flags
|= SQLITE_WriteSchema
| SQLITE_IgnoreChecks
| SQLITE_PreferBuiltin
;
147 db
->flags
&= ~(SQLITE_ForeignKeys
| SQLITE_ReverseOrder
);
150 pMain
= db
->aDb
[0].pBt
;
151 isMemDb
= sqlite3PagerIsMemdb(sqlite3BtreePager(pMain
));
153 /* Attach the temporary database as 'vacuum_db'. The synchronous pragma
154 ** can be set to 'off' for this file, as it is not recovered if a crash
155 ** occurs anyway. The integrity of the database is maintained by a
156 ** (possibly synchronous) transaction opened on the main database before
157 ** sqlite3BtreeCopyFile() is called.
159 ** An optimisation would be to use a non-journaled pager.
160 ** (Later:) I tried setting "PRAGMA vacuum_db.journal_mode=OFF" but
161 ** that actually made the VACUUM run slower. Very little journalling
162 ** actually occurs when doing a vacuum since the vacuum_db is initially
163 ** empty. Only the journal header is written. Apparently it takes more
164 ** time to parse and run the PRAGMA to turn journalling off than it does
165 ** to write the journal header file.
168 if( sqlite3TempInMemory(db
) ){
169 zSql
= "ATTACH ':memory:' AS vacuum_db;";
171 zSql
= "ATTACH '' AS vacuum_db;";
173 rc
= execSql(db
, pzErrMsg
, zSql
);
175 pDb
= &db
->aDb
[db
->nDb
-1];
176 assert( strcmp(pDb
->zName
,"vacuum_db")==0 );
178 if( rc
!=SQLITE_OK
) goto end_of_vacuum
;
179 pTemp
= db
->aDb
[db
->nDb
-1].pBt
;
181 /* The call to execSql() to attach the temp database has left the file
182 ** locked (as there was more than one active statement when the transaction
183 ** to read the schema was concluded. Unlock it here so that this doesn't
184 ** cause problems for the call to BtreeSetPageSize() below. */
185 sqlite3BtreeCommit(pTemp
);
187 nRes
= sqlite3BtreeGetReserve(pMain
);
189 /* A VACUUM cannot change the pagesize of an encrypted database. */
190 #ifdef SQLITE_HAS_CODEC
191 if( db
->nextPagesize
){
192 extern void sqlite3CodecGetKey(sqlite3
*, int, void**, int*);
195 sqlite3CodecGetKey(db
, 0, (void**)&zKey
, &nKey
);
196 if( nKey
) db
->nextPagesize
= 0;
200 rc
= execSql(db
, pzErrMsg
, "PRAGMA vacuum_db.synchronous=OFF");
201 if( rc
!=SQLITE_OK
) goto end_of_vacuum
;
203 /* Begin a transaction and take an exclusive lock on the main database
204 ** file. This is done before the sqlite3BtreeGetPageSize(pMain) call below,
205 ** to ensure that we do not try to change the page-size on a WAL database.
207 rc
= execSql(db
, pzErrMsg
, "BEGIN;");
208 if( rc
!=SQLITE_OK
) goto end_of_vacuum
;
209 rc
= sqlite3BtreeBeginTrans(pMain
, 2);
210 if( rc
!=SQLITE_OK
) goto end_of_vacuum
;
212 /* Do not attempt to change the page size for a WAL database */
213 if( sqlite3PagerGetJournalMode(sqlite3BtreePager(pMain
))
214 ==PAGER_JOURNALMODE_WAL
){
215 db
->nextPagesize
= 0;
218 if( sqlite3BtreeSetPageSize(pTemp
, sqlite3BtreeGetPageSize(pMain
), nRes
, 0)
219 || (!isMemDb
&& sqlite3BtreeSetPageSize(pTemp
, db
->nextPagesize
, nRes
, 0))
220 || NEVER(db
->mallocFailed
)
226 #ifndef SQLITE_OMIT_AUTOVACUUM
227 sqlite3BtreeSetAutoVacuum(pTemp
, db
->nextAutovac
>=0 ? db
->nextAutovac
:
228 sqlite3BtreeGetAutoVacuum(pMain
));
231 /* Query the schema of the main database. Create a mirror schema
232 ** in the temporary database.
234 rc
= execExecSql(db
, pzErrMsg
,
235 "SELECT 'CREATE TABLE vacuum_db.' || substr(sql,14) "
236 " FROM sqlite_master WHERE type='table' AND name!='sqlite_sequence'"
237 " AND coalesce(rootpage,1)>0"
239 if( rc
!=SQLITE_OK
) goto end_of_vacuum
;
240 rc
= execExecSql(db
, pzErrMsg
,
241 "SELECT 'CREATE INDEX vacuum_db.' || substr(sql,14)"
242 " FROM sqlite_master WHERE sql LIKE 'CREATE INDEX %' ");
243 if( rc
!=SQLITE_OK
) goto end_of_vacuum
;
244 rc
= execExecSql(db
, pzErrMsg
,
245 "SELECT 'CREATE UNIQUE INDEX vacuum_db.' || substr(sql,21) "
246 " FROM sqlite_master WHERE sql LIKE 'CREATE UNIQUE INDEX %'");
247 if( rc
!=SQLITE_OK
) goto end_of_vacuum
;
249 /* Loop through the tables in the main database. For each, do
250 ** an "INSERT INTO vacuum_db.xxx SELECT * FROM main.xxx;" to copy
251 ** the contents to the temporary database.
253 rc
= execExecSql(db
, pzErrMsg
,
254 "SELECT 'INSERT INTO vacuum_db.' || quote(name) "
255 "|| ' SELECT * FROM main.' || quote(name) || ';'"
256 "FROM main.sqlite_master "
257 "WHERE type = 'table' AND name!='sqlite_sequence' "
258 " AND coalesce(rootpage,1)>0"
260 if( rc
!=SQLITE_OK
) goto end_of_vacuum
;
262 /* Copy over the sequence table
264 rc
= execExecSql(db
, pzErrMsg
,
265 "SELECT 'DELETE FROM vacuum_db.' || quote(name) || ';' "
266 "FROM vacuum_db.sqlite_master WHERE name='sqlite_sequence' "
268 if( rc
!=SQLITE_OK
) goto end_of_vacuum
;
269 rc
= execExecSql(db
, pzErrMsg
,
270 "SELECT 'INSERT INTO vacuum_db.' || quote(name) "
271 "|| ' SELECT * FROM main.' || quote(name) || ';' "
272 "FROM vacuum_db.sqlite_master WHERE name=='sqlite_sequence';"
274 if( rc
!=SQLITE_OK
) goto end_of_vacuum
;
277 /* Copy the triggers, views, and virtual tables from the main database
278 ** over to the temporary database. None of these objects has any
279 ** associated storage, so all we have to do is copy their entries
280 ** from the SQLITE_MASTER table.
282 rc
= execSql(db
, pzErrMsg
,
283 "INSERT INTO vacuum_db.sqlite_master "
284 " SELECT type, name, tbl_name, rootpage, sql"
285 " FROM main.sqlite_master"
286 " WHERE type='view' OR type='trigger'"
287 " OR (type='table' AND rootpage=0)"
289 if( rc
) goto end_of_vacuum
;
291 /* At this point, there is a write transaction open on both the
292 ** vacuum database and the main database. Assuming no error occurs,
293 ** both transactions are closed by this block - the main database
294 ** transaction by sqlite3BtreeCopyFile() and the other by an explicit
295 ** call to sqlite3BtreeCommit().
301 /* This array determines which meta meta values are preserved in the
302 ** vacuum. Even entries are the meta value number and odd entries
303 ** are an increment to apply to the meta value after the vacuum.
304 ** The increment is used to increase the schema cookie so that other
305 ** connections to the same database will know to reread the schema.
307 static const unsigned char aCopy
[] = {
308 BTREE_SCHEMA_VERSION
, 1, /* Add one to the old schema cookie */
309 BTREE_DEFAULT_CACHE_SIZE
, 0, /* Preserve the default page cache size */
310 BTREE_TEXT_ENCODING
, 0, /* Preserve the text encoding */
311 BTREE_USER_VERSION
, 0, /* Preserve the user version */
312 BTREE_APPLICATION_ID
, 0, /* Preserve the application id */
315 assert( 1==sqlite3BtreeIsInTrans(pTemp
) );
316 assert( 1==sqlite3BtreeIsInTrans(pMain
) );
318 /* Copy Btree meta values */
319 for(i
=0; i
<ArraySize(aCopy
); i
+=2){
320 /* GetMeta() and UpdateMeta() cannot fail in this context because
321 ** we already have page 1 loaded into cache and marked dirty. */
322 sqlite3BtreeGetMeta(pMain
, aCopy
[i
], &meta
);
323 rc
= sqlite3BtreeUpdateMeta(pTemp
, aCopy
[i
], meta
+aCopy
[i
+1]);
324 if( NEVER(rc
!=SQLITE_OK
) ) goto end_of_vacuum
;
327 rc
= sqlite3BtreeCopyFile(pMain
, pTemp
);
328 if( rc
!=SQLITE_OK
) goto end_of_vacuum
;
329 rc
= sqlite3BtreeCommit(pTemp
);
330 if( rc
!=SQLITE_OK
) goto end_of_vacuum
;
331 #ifndef SQLITE_OMIT_AUTOVACUUM
332 sqlite3BtreeSetAutoVacuum(pMain
, sqlite3BtreeGetAutoVacuum(pTemp
));
336 assert( rc
==SQLITE_OK
);
337 rc
= sqlite3BtreeSetPageSize(pMain
, sqlite3BtreeGetPageSize(pTemp
), nRes
,1);
340 /* Restore the original value of db->flags */
341 db
->flags
= saved_flags
;
342 db
->nChange
= saved_nChange
;
343 db
->nTotalChange
= saved_nTotalChange
;
344 db
->xTrace
= saved_xTrace
;
345 sqlite3BtreeSetPageSize(pMain
, -1, -1, 1);
347 /* Currently there is an SQL level transaction open on the vacuum
348 ** database. No locks are held on any other files (since the main file
349 ** was committed at the btree level). So it safe to end the transaction
350 ** by manually setting the autoCommit flag to true and detaching the
351 ** vacuum database. The vacuum_db journal file is deleted when the pager
352 ** is closed by the DETACH.
357 sqlite3BtreeClose(pDb
->pBt
);
362 /* This both clears the schemas and reduces the size of the db->aDb[]
364 sqlite3ResetAllSchemasOfConnection(db
);
369 #endif /* SQLITE_OMIT_VACUUM && SQLITE_OMIT_ATTACH */