mksqlite documentation

mksqlite is an open source project to implement an interface between MATLAB(R) and SQLite. You can use the command mksqlite to benefit from all features of SQLite.

Simply enter the command mksqlite('SELECT * FROM database') and you get an array of structures with the results of the query.

For a more detailed documentation visit the project home page: http://mksqlite.sourceforge.net.

Why using mksqlite?

Since MATLAB(R) version 2016a the is a sqlite(...) function for easy access to SQLite databases.
mksqlite on the other hand serves the facility to...

• Handle BLOB data type (uint8 array)
• Save numerical arrays as (compressed) BLOB (shadowed)
• Save complex data structures (struct, cell, ...) as BLOB (shadowed)
• Define your own SQLite functions in terms of MATLAB(R) function callbacks Additionally mksqlite has some nice features like for example bulk insertion and named parameter binding:

clear s
s(1).Name = 'Triangle';
s(1).Type = 'Shape';
s(1).ID   = 145;
s(2).Name = 'Rectangle';
s(2).Type = 'Shape';
s(2).ID   = 344;

mksqlite( 'BEGIN' );
mksqlite( 'INSERT INTO tbl (ID, Type, name) VALUES (:ID, :Type; :Name)', s );
mksqlite( 'COMMIT' );

Prerequisites

mksqlite is build on top of SQLite and tested on various MATLAB(R) versions (R13SP1, R14SP2, 2006a, 2008b, 2009a, 2010a..2018b). You can use mksqlite on Windows, Linux and MAC OS-X systems.
The precompiled windows binaries are available from https://sourceforge.net/projects/mksqlite/files.
Unzip the contents of the ZIP file into a directory and insert this directory to the MATLAB search path.

addpath('path-to-mksqlite');

You can create other versions of mksqlite via the source package.

Using mksqlite source code

You can find the sourcecode of mksqlite also at https://sourceforge.net/projects/mksqlite/files. The Microsoft Visual C++ Compiler version 2003, 2005, 2010..2017 is needed to compile mksqlite on windows systems. Just use the MATLAB(R) script "buildit.m" to compile them to a MEX-DLL.

Online help

There is a small help available, just type help mksqlite in the MATLAB(R) command window.
For detailed description and lots of samples watch the compiled help file (.CHM) in the /doc folder.

Usage

You can use mksqlite like any other M-Script command:

dbid = mksqlite([dbid, ] SQLQuery [, Argument])

The parameter dbid is optional and is needed only if you want to use several data bases at the same time. If dbid was omitted, then the data base id 1 is used. Function calls:

mksqlite('open', 'databasefile' [,'ro'|'rw'|'rwc'] [,'single'|'multi'|'serial'])

or

dbid = mksqlite(0, 'open', 'databasefile' [,...])

Opens the database "databasefile". If there is no such database, then it will be created. If you use the dbid of an already opened database, than this db will be closed first. If you use the dbid 0, then the next free dbid will be used and returned by the function call.

mksqlite('close')

or

mksqlite(dbid, 'close')

or

mksqlite(0, 'close')

Closes the databasefile. If you use dbid 0, then all databases will be closed.

mksqlite('SQL-Query')

or

mksqlite(dbid, 'SQL-Query')

Executes the query and returns the results..

FAQ

My Windows MATLAB(R) doesn't find mksqlite.

Be sure that you have placed mksqlite.mex* inside the MATLAB(R) path or in the current directory.

MATLAB(R) shows "Attempt to execute SCRIPT mksqlite as a function"

It seems that you are using an older version of MATLAB(R). Try to rename the file "mksqlite.mexw32" to "mksqlite.dll" or build a new version of mksqlite from the source package.

Does mksqlite work on 64Bit MATLAB(R), on Solaris, Linux, OS-X, ...

Yes, there are 64Bit versions inside the binary package. (mksqlite.mexw64, mksqlite.mexa64)

How can I rebuild mksqlite?

You need MATLAB(R) and a compiler. The compiler must be useable to build MATLAB(R) MEX-extensions. You can use on of the Microsoft Compiler on Windows, or a GNU compiler on other Systems. You will find the MATLAB(R) script "buildit.m" inside the source package. Just start it from the MATLAB(R) prompt.

How do I build mksqlite on MacOS? (Will build Intel version on Intel Macs and arm64 version on Macs running Apple Silicon.)

1. Install XCode from the App Store
2. Install XCode command line tools by typing xcode-select --install on the command line.
3. Be sure to accept the Xcode license by typing sudo xcodebuild -license
4. You will need to install CMake, if you do not have it already installed.
5. An easy way to do this is to install MacPorts.
6. Then, type sudo port install cmake or sudo /opt/local/bin/port install cmake
7. Edit your ~/.zshrc file to include the line path+=('/opt/local/bin') so that /opt/local/bin is on your path. This is where cmake was installed with MacPorts.
8. Follow the directions in CmakeLists.txt. You should not use the -G option with cmake or the Release configuration.
9. The file mksqlite.mexmaci64 or mksqlite.mexmaca64 will be in your build directory. Copy it to the mksqlite directory.

The SQLite database seems to be very slow. Can I speedup my operations?

SQLite uses a journal file for every write operation. This slows down any change on the database. Try to use transactions for a block of change operations or change the journal functions with SQLite pragma commands. You can find more informations inside of the SQLite documentation.

How can I catch mksqlite errors?

You can use the standard MATLAB(R) try ... catch functions.

What does mksqlite costs?

You can use mksqlite without any cost. But if you mean that this software is worth a donation, Just send me a mail with your offer.

I need a new functionality, what's about new functions?

If you have implemented a new function or have an idea how something can made better, just send me an email. Maybe I will implement this function or I can give you a hint how you can work with the current functionality.

Can I do something to help develop mksqlite?

Sure, documentation, translation, extensions, give me notes about errors, ... Or just give me a note how do you use mksqlite.

License

mksqlite is distributed under the BSD-2 license (see LICENSE file)

Contact

Andreas Martin, [email protected]
Martin Kortmann, [email protected]