Backup

A backup object encapsulates copying one database to another. You call Connection.backup() on the destination database to get the Backup object. Call step() to copy some pages repeatedly dealing with errors as appropriate. Finally finish() cleans up committing or rolling back and releasing locks.

Here is an example usage using the with statement to ensure finish() is called:

# copies source.main into db
with db.backup("main", source, "main") as b:
    while not b.done:
        b.step(100)
        print(b.remaining, b.pagecount, "\r", flush = True)

If you are not using with then you’ll need to ensure finish() is called:

# copies source.main into db
b=db.backup("main", source, "main")
try:
    while not b.done:
        b.step(100)
        print(b.remaining, b.pagecount, "\r", flush = True)
finally:
    b.finish()

Important details

The database is copied page by page. This means that there is not a round trip via SQL. All pages are copied including free ones.

The destination database is locked during the copy. You will get a ThreadingViolationError if you attempt to use it.

Backup class

class Backup

You create a backup instance by calling Connection.backup().

Backup.__enter__() → Backup

You can use the backup object as a context manager as defined in PEP 0343. The __exit__() method ensures that backup is finished.

Backup.__exit__(etype: Optional[type[BaseException]], evalue: Optional[BaseException], etraceback: Optional[types.TracebackType]) → Optional[bool]

Implements context manager in conjunction with __enter__() ensuring that the copy is finished.

Backup.close(force: bool = False) → None

Does the same thing as finish(). This extra api is provided to give the same api as other APSW objects such as Connection.close(), Blob.close() and Cursor.close(). It is safe to call this method multiple times.

Parameters

force – If true then any exceptions are ignored.

Backup.done: bool

A boolean that is True if the copy completed in the last call to step().

Backup.finish() → None

Completes the copy process. If all pages have been copied then the transaction is committed on the destination database, otherwise it is rolled back. This method must be called for your backup to take effect. The backup object will always be finished even if there is an exception. It is safe to call this method multiple times.

Calls: sqlite3_backup_finish

Backup.pagecount: int

Read only. How many pages were in the source database after the last step. If you haven’t called step() or the backup object has been finished then zero is returned.

Calls: sqlite3_backup_pagecount

Backup.remaining: int

Read only. How many pages were remaining to be copied after the last step. If you haven’t called step() or the backup object has been finished then zero is returned.

Calls: sqlite3_backup_remaining

Backup.step(npages: int = -1) → bool

Copies npages pages from the source to destination database. The source database is locked during the copy so using smaller values allows other access to the source database. The destination database is always locked until the backup object is finished.

Parameters

npages – How many pages to copy. If the parameter is omitted or negative then all remaining pages are copied. The default page size is 1024 bytes (1kb) which can be changed before database creation using a pragma.

This method may throw a BusyError or LockedError if unable to lock the source database. You can catch those and try again.

Returns

True if this copied the last remaining outstanding pages, else false. This is the same value as done

Calls: sqlite3_backup_step