Warning

This section contains snippets that were automatically translated from C++ to Python and may contain errors.

Executing SQL Statements

The QSqlQuery class provides an interface for executing SQL statements and navigating through the result set of a query.

The QSqlQueryModel and QSqlTableModel classes described in the next section provide a higher-level interface for accessing databases. If you are unfamiliar with SQL, you might want to skip directly to the next section ( Using the SQL Model Classes ).

Executing a Query

To execute an SQL statement, simply create a QSqlQuery object and call exec() like this:

query = QSqlQuery()
query.exec("SELECT name, salary FROM employee WHERE salary > 50000")

The QSqlQuery constructor accepts an optional QSqlDatabase object that specifies which database connection to use. In the example above, we don’t specify any connection, so the default connection is used.

If an error occurs, exec() returns false. The error is then available as lastError() .

Navigating the Result Set

QSqlQuery provides access to the result set one record at a time. After the call to exec() , QSqlQuery ‘s internal pointer is located one position before the first record. We must call next() once to advance to the first record, then next() again repeatedly to access the other records, until it returns false. Here’s a typical loop that iterates over all the records in order:

while query.next():
    name = query.value(0).toString()
    salary = query.value(1).toInt()
    print(name, salary)

The value() function returns the value of a field in the current record. Fields are specified as zero-based indexes. value() returns a QVariant , a type that can hold various C++ and core Qt data types such as int, QString , and QByteArray . The different database types are automatically mapped into the closest Qt equivalent. In the code snippet, we call toString() and toInt() to convert variants to QString and int.

For an overview of the recommended types for use with Qt-supported Databases, please refer to this table .

You can navigate within the dataset using next() , previous() , first() , last() , and seek() . The current row index is returned by at() , and the total number of rows in the result set is available as size() for databases that support it.

To determine whether a database driver supports a given feature, use hasFeature() . In the following example, we call size() to determine the size of a result set of the underlying database supports that feature; otherwise, we navigate to the last record and use the query’s position to tell us how many records there are.

query = QSqlQuery()
numRows = int()
query.exec("SELECT name, salary FROM employee WHERE salary > 50000")
defaultDB = QSqlDatabase.database()
if defaultDB.driver().hasFeature(QSqlDriver.QuerySize):
    numRows = query.size()
else:
    # this can be very slow
    query.last()
    numRows = query.at() + 1

If you navigate within a result set, and use next() and seek() only for browsing forward, you can call setForwardOnly (true) before calling exec(). This is an easy optimization that will speed up the query significantly when operating on large result sets.

Inserting, Updating, and Deleting Records

QSqlQuery can execute arbitrary SQL statements, not just SELECTs. The following example inserts a record into a table using INSERT:

query = QSqlQuery()
query.exec("INSERT INTO employee (id, name, salary) "
           "VALUES (1001, 'Thad Beaumont', 65000)")

If you want to insert many records at the same time, it is often more efficient to separate the query from the actual values being inserted. This can be done using placeholders. Qt supports two placeholder syntaxes: named binding and positional binding. Here’s an example of named binding:

query = QSqlQuery()
query.prepare("INSERT INTO employee (id, name, salary) "
              "VALUES (:id, :name, :salary)")
query.bindValue(":id", 1001)
query.bindValue(":name", "Thad Beaumont")
query.bindValue(":salary", 65000)
query.exec()

Here’s an example of positional binding:

query = QSqlQuery()
query.prepare("INSERT INTO employee (id, name, salary) "
              "VALUES (?, ?, ?)")
query.addBindValue(1001)
query.addBindValue("Thad Beaumont")
query.addBindValue(65000)
query.exec()

Both syntaxes work with all database drivers provided by Qt. If the database supports the syntax natively, Qt simply forwards the query to the DBMS; otherwise, Qt simulates the placeholder syntax by preprocessing the query. The actual query that ends up being executed by the DBMS is available as executedQuery() .

When inserting multiple records, you only need to call prepare() once. Then you call bindValue() or addBindValue() followed by exec() as many times as necessary.

Besides performance, one advantage of placeholders is that you can easily specify arbitrary values without having to worry about escaping special characters.

Updating a record is similar to inserting it into a table:

query = QSqlQuery()
query.exec("UPDATE employee SET salary = 70000 WHERE id = 1003")

You can also use named or positional binding to associate parameters to actual values.

Finally, here’s an example of a DELETE statement:

query = QSqlQuery()
query.exec("DELETE FROM employee WHERE id = 1007")

Transactions

If the underlying database engine supports transactions, hasFeature ( Transactions ) will return true. You can use transaction() to initiate a transaction, followed by the SQL commands you want to execute within the context of the transaction, and then either commit() or rollback() . When using transactions you must start the transaction before you create your query.

Example:

QSqlDatabase.database().transaction()
query = QSqlQuery()
query.exec("SELECT id FROM employee WHERE name = 'Torild Halvorsen'")
if query.next():
    employeeId = query.value(0).toInt()
    query.exec("INSERT INTO project (id, name, ownerid) "
               "VALUES (201, 'Manhattan Project', "
               + QString.number(employeeId) + ')')

QSqlDatabase.database().commit()

Transactions can be used to ensure that a complex operation is atomic (for example, looking up a foreign key and creating a record), or to provide a means of canceling a complex change in the middle.