Matlab driver for SQLite3 database
Switch branches/tags
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
+sqlite3
include
src
test
.gitignore
LICENSE
Makefile
README.md

README.md

Matlab SQLite3 Driver

Matlab driver for SQLite3 database. Features include:

  • Simple and clean Matlab API.
  • Fast C++ MEX implementation.
  • Parameter binding for an SQL statement.
  • Multiple database connections.
  • Easy manipulation of query results as a struct array.

Here is a quick example.

addpath('/path/to/matlab-sqlite3-driver');
sqlite3.open('/path/to/database.sqlite3');
sqlite3.execute('CREATE TABLE records (id INTEGER, name VARCHAR)');
sqlite3.execute('INSERT INTO records VALUES (?, ?)', 1, 'foo');
sqlite3.execute('INSERT INTO records VALUES (?, ?)', 2, 'bar');
records = sqlite3.execute('SELECT * FROM records WHERE id < ?', 10);
result = sqlite3.execute('SELECT COUNT(*) FROM records');
sqlite3.close();

Prerequisites

This driver requires Matlab R2011b or later.

This driver is built on the Matlab mex interface. The package requires a C++ compiler for mex. In Windows, a C++ compiler is available from Visual Studio and Microsoft SDK. Check Mathworks webpage for a supported compiler.

In Linux, a compiler is usually available in a package manager. For example, in Debian/Ubuntu:

$ apt-get install build-essential

In Mac OS, install XCode and also UNIX command line tools. Also check this page to apply patches if necessary.

After installing build tools, set up the mex command in Matlab.

>> mex -setup

Build

Add path to the matlab-sqlite3-driver first. Then, call sqlite3.make in Matlab.

>> sqlite3.make

Runtime requirement

Always add path to this driver before use within Matlab.

>> addpath /path/to/matlab-sqlite3-driver

In most cases, you'll need to force Matlab to preload the system library due to the incompatibility between Matlab's internal C++ runtime. Use LD_PRELOAD variable to do so. For example, in Ubuntu 12.04 LTS 64-bit,

LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libstdc++.so.6:/lib/x86_64-linux-gnu/libgcc_s.so.1 matlab

The required path depends on the OS. You can check which path to specify by comparing the output of ldd tool in the UNIX shell and in Matlab.

From the UNIX shell,

$ ldd +sqlite3/private/libsqlite3_.mex*

From Matlab shell,

>> !ldd +sqlite3/private/libsqlite3_.mex*

And find a dependent library differing in Matlab. You must append that library in the LD_PRELOAD path.

In OS X, LD_PRELOAD equivalent is DYLD_INSERT_LIBRARIES. Use otool -L command instead of ldd.

Test

Optionally test the functionality with the attached test script.

>> addpath test/
>> test_sqlite3

API

There are 4 public functions. All functions are scoped under sqlite3 namespace. Also check help of each function.

open      Open a database.
close     Close a database connection.
execute   Execute an SQLite statement.
timeout   Set timeout value when database is busy.

open

database = sqlite3.open(filename)

The open operation takes a file name of a database and returns newly created connection id. This id can be used for operations until closed.

Example:

>> database = sqlite3.open('/path/to/test.db');

close

sqlite3.close(database)
sqlite3.close()

The close operation closes connection to the database specified by the connection id database. When database is omitted, the last opened connection is closed.

execute

results = sqlite3.execute(database, sql, param1, param2, ...)
results = sqlite3.execute(sql, param1, param2, ...)

The execute operation applies a sql statement sql in a database specified by the connection id database. When database is omitted, the last opened connection is used.

The sql statement can bind parameters through ? as a placeholder. When binding is used, there must be corresponding number of parameters following the sql statement. Bind values can be a numeric scalar value, a string, a uint8 array for blob, or an empty array for null.

Results are returned as a struct array.

Example:

>> results = sqlite3.execute(database, 'SELECT * FROM records');
>> results = sqlite3.execute('SELECT * FROM records');
>> results = sqlite3.execute('SELECT * FROM records WHERE rowid = ? OR name = ?', 1, 'foo');
>> results = sqlite3.execute('INSERT INTO records VALUES (?)', 'bar');

Metadata can be retrieved from sqlite_master table or from PRAGMA statement.

>> tables = sqlite3.execute('SELECT * FROM sqlite_master WHERE type="table"');
>> indices = sqlite3.execute('SELECT * FROM sqlite_master WHERE type="index"');
>> columns = sqlite3.execute('PRAGMA TABLE_INFO(records)');

timeout

sqlite3.timeout(database, millisecond)
sqlite3.timeout(millisecond)

The timeout operation sets how long the driver should wait when the database is locked by other processes.

Example:

>> sqlite3.timeout(1000);

Tips

Mass insertion

Use transaction to insert a lot of data.

sqlite3.execute('BEGIN');
for i = 1:numel(X)
    sqlite3.execute('INSERT INTO records (x) values (?)', X(i));
end
sqlite3.execute('COMMIT');

Or, convert an array to a statement string.

values = sprintf('(%g),', X);
sqlite3.execute(['INSERT INTO records (x) values ', values(1:end-1)]);

License

The code may be redistributed under BSD 3-clause license.