The London Perl and Raku Workshop takes place on 26th Oct 2024. If your company depends on Perl, please consider sponsoring and/or attending.


DBIx::SearchBuilder::Handle - Perl extension which is a generic DBI handle


  use DBIx::SearchBuilder::Handle;

  my $handle = DBIx::SearchBuilder::Handle->new();
  $handle->Connect( Driver => 'mysql',
                    Database => 'dbname',
                    Host => 'hostname',
                    User => 'dbuser',
                    Password => 'dbpassword');
  # now $handle isa DBIx::SearchBuilder::Handle::mysql


This class provides a wrapper for DBI handles that can also perform a number of additional functions.


Generic constructor

Connect PARAMHASH: Driver, Database, Host, User, Password, QuoteTableNames

Takes a paramhash and connects to your DBI datasource.

You should _always_ set

     DisconnectHandleOnDestroy => 1

unless you have a legacy app like RT2 or RT 3.0.{0,1,2} that depends on the broken behaviour.

If you created the handle with DBIx::SearchBuilder::Handle->new and there is a DBIx::SearchBuilder::Handle::(Driver) subclass for the driver you have chosen, the handle will be automatically "upgraded" into that subclass.

QuoteTableNames option will force all table names to be quoted if the driver subclass has a method for quoting implemented. The mysql subclass will detect mysql version 8 and set the flag.

_UpgradeHandle DRIVER

This private internal method turns a plain DBIx::SearchBuilder::Handle into one of the standard driver-specific subclasses.


Takes a bunch of parameters:

Required: Driver, Database, Optional: Host, Port and RequireSSL

Builds a DSN suitable for a DBI connection


Returns the DSN for this database connection.

RaiseError [MODE]

Turns on the Database Handle's RaiseError attribute.

PrintError [MODE]

Turns on the Database Handle's PrintError attribute.

LogSQLStatements BOOL

Takes a boolean argument. If the boolean is true, SearchBuilder will log all SQL statements, as well as their invocation times and execution times.

Returns whether we're currently logging or not as a boolean


Add an SQL statement to our query log


Clears out the SQL statement log.


Returns the current SQL statement log as an array of arrays. Each entry is a triple of

(Time, Statement, Duration)

AutoCommit [MODE]

Turns on the Database Handle's AutoCommit attribute.


Disconnect from your DBI datasource

dbh [HANDLE]

Return the current DBI handle. If we're handed a parameter, make the database handle that.


Takes a table name and a set of key-value pairs in an array. Splits the key value pairs, constructs an INSERT statement and performs the insert.

Base class return statement handle object, while DB specific subclass should return row id.


Takes a table name and a set of key-value pairs in an array. Splits the key value pairs, constructs an INSERT statement and returns query string and set of bind values.

This method is more useful for subclassing in DB specific handles. "Insert" method is preferred for end users.


Takes table name, array reference with columns, select query and list of bind values. Inserts data select by the query into the table.

To make sure call is portable every column in result of the query should have unique name or should be aliased. See "InsertFromSelect" in DBIx::SearchBuilder::Handle::Oracle for details.


Takes a hash with fields: Table, Column, Value PrimaryKeys, and IsSQLFunction. Table, and Column should be obvious, Value is where you set the new value you want the column to have. The primary_keys field should be the lvalue of DBIx::SearchBuilder::Record::PrimaryKeys(). Finally IsSQLFunction is set when the Value is a SQL function. For example, you might have ('Value'=>'PASSWORD(string)'), by setting IsSQLFunction that string will be inserted into the query directly rather then as a binding.


Update column COLUMN of table TABLE where the record id = RECORD_ID. if IS_SQL is set, don\'t quote the NEW_VALUE


Takes table name, hash reference with (column, value) pairs, select query and list of bind values.

Updates the table, but only records with IDs returned by the selected query, eg:

    UPDATE $table SET %values WHERE id IN ( $query )

It's simple as values are static and search only allowed by id.


Takes table name, select query and list of bind values.

Deletes from the table, but only records with IDs returned by the select query, eg:

    DELETE FROM $table WHERE id IN ($query)

SimpleQuery QUERY_STRING, [ BIND_VALUE, ... ]

Execute the SQL string specified in QUERY_STRING

FetchResult QUERY, [ BIND_VALUE, ... ]

Takes a SELECT query as a string, along with an array of BIND_VALUEs If the select succeeds, returns the first row as an array. Otherwise, returns a Class::ResturnValue object with the failure loaded up.


Returns 1 if the current database supports BLOBs with embedded nulls. Returns undef if the current database doesn't support BLOBs with embedded nulls


Returns 1 if the current database supports inserts of BLOBs automatically. Returns undef if the current database must be informed of BLOBs for inserts.


Returns a hash ref for the bind_param call to identify BLOB types used by the current database for a particular column type.

DatabaseVersion [Short => 1]

Returns the database's version.

If argument Short is true returns short variant, in other case returns whatever database handle/driver returns. By default returns short version, e.g. '4.1.23' or '8.0-rc4'.

Returns empty string on error or if database couldn't return version.

The base implementation uses a SELECT VERSION()


Returns 1 if the current database's searches are case sensitive by default Returns undef otherwise


Returns 1 if table names will be quoted in queries, otherwise 0

_MakeClauseCaseInsensitive FIELD OPERATOR VALUE

Takes a field, operator and value. performs the magic necessary to make your database treat this clause as case insensitive.

Returns a FIELD OPERATOR VALUE triple.


DBIx::SearchBuilder::Handle emulates nested transactions, by keeping a transaction stack depth.

NOTE: In nested transactions you shouldn't mix rollbacks and commits, because only last action really do commit/rollback. For example next code would produce desired results:


Only last action(Commit in example) finilize transaction in DB.


Tells DBIx::SearchBuilder to begin a new SQL transaction. This will temporarily suspend Autocommit mode.

EndTransaction [Action => 'commit'] [Force => 0]

Tells to end the current transaction. Takes Action argument that could be commit or rollback, the default value is commit.

If Force argument is true then all nested transactions would be committed or rolled back.

If there is no transaction in progress then method throw warning unless action is forced.

Method returns true on success or false if an error occurred.

Commit [FORCE]

Tells to commit the current SQL transaction.

Method uses EndTransaction method, read its description.

Rollback [FORCE]

Tells to abort the current SQL transaction.

Method uses EndTransaction method, read its description.


Force the handle to rollback. Whether or not we're deep in nested transactions.


Returns the current depth of the nested transaction stack. Returns undef if there is no connection to database.


takes an SQL SELECT statement and massages it to return ROWS_PER_PAGE starting with FIRST_ROW;

Join { Paramhash }

Takes a paramhash of everything Searchbuildler::Record does plus a parameter called 'SearchBuilder' that contains a ref to a SearchBuilder object'.

This performs the join.


Takes a SearchBuilder and ALIAS in a hash and resturns true if restrictions of the query allow NULLs in a table joined with the ALIAS, otherwise returns false value which means that you can use normal join instead of left for the aliased table.

Works only for queries have been built with "Join" in DBIx::SearchBuilder and "Limit" in DBIx::SearchBuilder methods, for other cases return true value to avoid fault optimizations.


takes an incomplete SQL SELECT statement and massages it to return a DISTINCT result set.

DistinctQueryAndCount STATEMENTREF

takes an incomplete SQL SELECT statement and massages it to return a DISTINCT result set and the total count of potential records.


takes an incomplete SQL SELECT statement and massages it to return a DISTINCT result set.


Takes a single argument, a message to log.

Currently prints that message to STDERR


See "DateTimeFunction" for details on supported functions. This method is for implementers of custom DB connectors.

Returns hash reference with (function name, sql template) pairs.


Takes named arguments:

  • Field - SQL expression date/time function should be applied to. Note that this argument is used as is without any kind of quoting.

  • Type - name of the function, see supported values below.

  • Timezone - optional hash reference with From and To values, see "ConvertTimezoneFunction" for details.

Returns SQL statement. Returns NULL if function is not supported.

Supported functions

Type value in "DateTimeFunction" is case insesitive. Spaces, underscores and dashes are ignored. So 'date time', 'DateTime' and 'date_time' are all synonyms. The following functions are supported:

  • date time - as is, no conversion, except applying timezone conversion if it's provided.

  • time - time only

  • hourly - datetime prefix up to the hours, e.g. '2010-03-25 16'

  • hour - hour, 0 - 23

  • date - date only

  • daily - synonym for date

  • day of week - 0 - 6, 0 - Sunday

  • day - day of month, 1 - 31

  • day of month - synonym for day

  • day of year - 1 - 366, support is database dependent

  • month - 1 - 12

  • monthly - year and month prefix, e.g. '2010-11'

  • year - e.g. '2023'

  • annually - synonym for year

  • week of year - 0-53, presence of zero week, 1st week meaning and whether week starts on Monday or Sunday heavily depends on database.


Generates a function applied to Field argument that converts timezone. By default converts from UTC. Examples:

    # UTC => Moscow
    $handle->ConvertTimezoneFunction( Field => '?', To => 'Europe/Moscow');

If there is problem with arguments or timezones are equal then Field returned without any function applied. Field argument is not escaped in any way, it's your job.

Implementation is very database specific. To be portable convert from UTC or to UTC. Some databases have internal storage for information about timezones that should be kept up to date. Read documentation for your DB.


Generates a function to calculate interval in seconds between two dates. Takes From and To arguments which can be either scalar or a hash. Hash is processed with "CombineFunctionWithField" in DBIx::SearchBuilder.

Arguments are not quoted or escaped in any way. It's caller's job.


Sets order of NULLs when sorting columns when called with mode, but only if DB supports it. Modes:

  • small

    NULLs are smaller then anything else, so come first when order is ASC and last otherwise.

  • large

    NULLs are larger then anything else.

  • first

    NULLs are always first.

  • last

    NULLs are always last.

  • default

    Return back to DB's default behaviour.

When called without argument returns metadata required to generate SQL.


Returns true value if DB supports adjusting NULLs order while sorting a column, for example ORDER BY Value ASC NULLS FIRST.


Returns true value if DB supports to combine search and count in single query.


Returns true value if DB supports empty string.


Quote table or column name to avoid reserved word errors.

Returns same value passed unless over-ridden in database-specific subclass.


Undo the effects of QuoteName by removing quoting.


When we get rid of the Searchbuilder::Handle, we need to disconnect from the database

CastAsDecimal FIELD

Cast the given field as decimal.

E.g. on Pg, it's CAST(Content AS DECIMAL).