PG::

Connection class

Superclass	rb_cObject
Included Modules	PG::Constants PG::Connection::Pollable

The PostgreSQL connection class. The interface for this class is based on libpq, the C application programmer’s interface to PostgreSQL. Some familiarity with libpq is recommended, but not necessary.

For example, to send query to the database on the localhost:

require 'pg'
conn = PG::Connection.open(:dbname => 'test')
res = conn.exec_params('SELECT $1 AS a, $2 AS b, $3 AS c', [1, 2, nil])
# Equivalent to:
#  res  = conn.exec('SELECT 1 AS a, 2 AS b, NULL AS c')

See the PG::Result class for information on working with the results of a query.

Many methods of this class have three variants kind of:

exec - the base method which is an alias to async_exec . This is the method that should be used in general.
async_exec - the async aware version of the method, implemented by libpq’s async API.
sync_exec - the method version that is implemented by blocking function(s) of libpq.

Sync and async version of the method can be switched by Connection.async_api= , however it is not recommended to change the default.

Public Class Methods

async_api=(enable)

Switch between sync and async libpq API.

PG::Connection.async_api = true

this is the default. It sets an alias from exec to async_exec, reset to async_reset and so on.

PG::Connection.async_api = false

sets an alias from exec to sync_exec, reset to sync_reset and so on.

pg-1.1.0+ defaults to libpq’s async API for query related blocking methods. pg-1.3.0+ defaults to libpq’s async API for all possibly blocking methods.

PLEASE NOTE: This method is not part of the public API and is for debug and development use only. Do not use this method in production code. Any issues with the default setting of async_api=true should be reported to the maintainers instead.

# File lib/pg/connection.rb, line 1079
def async_api=(enable)
        self.async_send_api = enable
        REDIRECT_METHODS.each do |ali, (async, sync)|
                remove_method(ali) if method_defined?(ali)
                alias_method( ali, enable ? async : sync )
        end
        REDIRECT_CLASS_METHODS.each do |ali, (async, sync)|
                singleton_class.remove_method(ali) if method_defined?(ali)
                singleton_class.alias_method(ali, enable ? async : sync )
        end
end

PG::Connection.new → conn
PG::Connection.new(connection_hash) → conn
PG::Connection.new(connection_string) → conn
PG::Connection.new(host, port, options, tty, dbname, user, password) → conn

Alias for: new

PG::Connection.ping(connection_hash) → Integer
PG::Connection.ping(connection_string) → Integer
PG::Connection.ping(host, port, options, tty, dbname, login, password) → Integer

Alias for: ping

async_send_api=(enable)

# File lib/pg/connection.rb, line 1056
def async_send_api=(enable)
        REDIRECT_SEND_METHODS.each do |ali, (async, sync)|
                undef_method(ali) if method_defined?(ali)
                alias_method( ali, enable ? async : sync )
        end
end

PG::Connection.conndefaults() → Array

Returns an array of hashes. Each hash has the keys:

:keyword

the name of the option

:envvar

the environment variable to fall back to

:compiled

the compiled in option as a secondary fallback

:val

the option’s current value, or nil if not known

:label

the label for the field

:dispchar

“” for normal, “D” for debug, and “*” for password

:dispsize

field size

static VALUE
pgconn_s_conndefaults(VALUE self)
{
        PQconninfoOption *options = PQconndefaults();
        VALUE array = pgconn_make_conninfo_array( options );

        PQconninfoFree(options);

        UNUSED( self );

        return array;
}

conndefaults_hash()

Return the Postgres connection defaults structure as a Hash keyed by option keyword (as a Symbol).

Create a connection to the specified server.

connection_hash must be a ruby Hash with connection parameters. See the list of valid parameters in the PostgreSQL documentation.

There are two accepted formats for connection_string: plain keyword = value strings and URIs. See the documentation of connection strings.

The positional parameter form has the same functionality except that the missing parameters will always take on default values. The parameters are:

host

server hostname

port

server port number

options

backend options

tty

(ignored in all versions of PostgreSQL)

dbname

connecting database name

user

password

If the Ruby default internal encoding is set (i.e., Encoding.default_internal != nil), the connection will have its client_encoding set accordingly.

Raises a PG::Error if the connection fails.

Examples:

# Connect using all defaults
PG::Connection.new

# As a Hash
PG::Connection.new( dbname: 'test', port: 5432 )

# As a String
PG::Connection.new( "dbname=test port=5432" )

# As an Array
PG::Connection.new( nil, 5432, nil, nil, 'test', nil, nil )

# As an URI
PG::Connection.new( "postgresql://user:pass@pgsql.example.com:5432/testdb?sslmode=require" )

Specifying Multiple Hosts

It is possible to specify multiple hosts to connect to, so that they are tried in the given order or optionally in random order. In the Keyword/Value format, the host, hostaddr, and port options accept comma-separated lists of values. The details to libpq describe how it works, but there are two small differences how ruby-pg handles multiple hosts:

All hosts are resolved before the first connection is tried. This means that when load_balance_hosts is set to random, then all resolved addresses are tried randomly in one level. When a host resolves to more than one address, it is therefore tried more often than a host that has only one address.
When a timeout occurs due to the value of connect_timeout, then the given host, hostaddr and port combination is not tried a second time, even if it’s specified several times. It’s still possible to do load balancing with load_balance_hosts set to random and to increase the number of connections a node gets, when the hostname is provided multiple times in the host string. This is because in non-timeout cases the host is tried multiple times.

# File lib/pg/connection.rb, line 869
def new(*args)
        conn = connect_to_hosts(*args)

        if block_given?
                begin
                        return yield conn
                ensure
                        conn.finish
                end
        end
        conn
end

Also aliased as: async_connect , connect , open , setdb , setdblogin

PG::Connection.new → conn
PG::Connection.new(connection_hash) → conn
PG::Connection.new(connection_string) → conn
PG::Connection.new(host, port, options, tty, dbname, user, password) → conn

Alias for: new

parse_connect_args( *args )

Parse the connection args into a connection-parameter string. See PG::Connection.new for valid arguments.

It accepts:

an option String kind of “host=name port=5432”
an option Hash kind of {host: “name”, port: 5432}
URI string
URI object
positional arguments

The method adds the option “fallback_application_name” if it isn’t already set. It returns a connection string with “key=value” pairs.

# File lib/pg/connection.rb, line 63
def self.parse_connect_args( *args )
        hash_arg = args.last.is_a?( Hash ) ? args.pop.transform_keys(&:to_sym) : {}
        iopts = {}

        if args.length == 1
                case args.first.to_s
                when /=/, /:\/\//
                        # Option or URL string style
                        conn_string = args.first.to_s
                        iopts = PG::Connection.conninfo_parse(conn_string).each_with_object({}){|h, o| o[h[:keyword].to_sym] = h[:val] if h[:val] }
                else
                        # Positional parameters (only host given)
                        iopts[CONNECT_ARGUMENT_ORDER.first.to_sym] = args.first
                end
        else
                # Positional parameters with host and more
                max = CONNECT_ARGUMENT_ORDER.length
                raise ArgumentError,
                                "Extra positional parameter %d: %p" % [ max + 1, args[max] ] if args.length > max

                CONNECT_ARGUMENT_ORDER.zip( args ) do |(k,v)|
                        iopts[ k.to_sym ] = v if v
                end
                iopts.delete(:tty) # ignore obsolete tty parameter
        end

        iopts.merge!( hash_arg )

        if !iopts[:fallback_application_name]
                iopts[:fallback_application_name] = PROGRAM_NAME.sub( /^(.{30}).{4,}(.{30})$/ ){ $1+"..."+$2 }
        end

        return connect_hash_to_string(iopts)
end

PG::Connection.ping(connection_hash) → Integer
PG::Connection.ping(connection_string) → Integer
PG::Connection.ping(host, port, options, tty, dbname, login, password) → Integer

PQpingParams reports the status of the server.

It accepts connection parameters identical to those of PQ::Connection.new . It is not necessary to supply correct user name, password, or database name values to obtain the server status; however, if incorrect values are provided, the server will log a failed connection attempt.

See PG::Connection.new for a description of the parameters.

Returns one of:

PQPING_OK

server is accepting connections

PQPING_REJECT

server is alive but rejecting connections

PQPING_NO_RESPONSE

could not establish connection

PQPING_NO_ATTEMPT

connection not attempted (bad params)

See also check_socket for a way to check the connection without doing any server communication.

# File lib/pg/connection.rb, line 984
def ping(*args)
        if Fiber.respond_to?(:scheduler) && Fiber.scheduler
                # Run PQping in a second thread to avoid blocking of the scheduler.
                # Unfortunately there's no nonblocking way to run ping.
                Thread.new { sync_ping(*args) }.value
        else
                sync_ping(*args)
        end
end

Also aliased as: async_ping

quote_connstr( value )

Quote a single value for use in a connection-parameter string.

# File lib/pg/connection.rb, line 36
def self.quote_connstr( value )
        return "'" + value.to_s.gsub( /[\\']/ ) {|m| '\\' + m } + "'"
end

quote_ident( str ) → String
quote_ident( array ) → String
PG::Connection.quote_ident( str ) → String
PG::Connection.quote_ident( array ) → String

Returns a string that is safe for inclusion in a SQL query as an identifier. Note: this is not a quote function for values, but for identifiers.

For example, in a typical SQL query: SELECT FOO FROM MYTABLE The identifier FOO is folded to lower case, so it actually means foo. If you really want to access the case-sensitive field name FOO, use this function like conn.quote_ident('FOO'), which will return "FOO" (with double-quotes). PostgreSQL will see the double-quotes, and it will not fold to lower case.

Similarly, this function also protects against special characters, and other things that might allow SQL injection if the identifier comes from an untrusted source.

If the parameter is an Array, then all it’s values are separately quoted and then joined by a “.” character. This can be used for identifiers in the form “schema”.“table”.“column” .

This method is functional identical to the encoder PG::TextEncoder::Identifier .

If the instance method form is used and the input string character encoding is different to the connection encoding, then the string is converted to this encoding, so that the returned string is always encoded as PG::Connection#internal_encoding .

In the singleton form (PG::Connection.quote_ident) the character encoding of the result string is set to the character encoding of the input string.

static VALUE
pgconn_s_quote_ident(VALUE self, VALUE str_or_array)
{
        VALUE ret;
        int enc_idx;

        if( rb_obj_is_kind_of(self, rb_cPGconn) ){
                enc_idx = pg_get_connection(self)->enc_idx;
        }else{
                enc_idx = RB_TYPE_P(str_or_array, T_STRING) ? ENCODING_GET( str_or_array ) : rb_ascii8bit_encindex();
        }
        pg_text_enc_identifier(NULL, str_or_array, NULL, &ret, enc_idx);

        return ret;
}

PG::Connection.new → conn
PG::Connection.new(connection_hash) → conn
PG::Connection.new(connection_string) → conn
PG::Connection.new(host, port, options, tty, dbname, user, password) → conn

Alias for: new

PG::Connection.new → conn
PG::Connection.new(connection_hash) → conn
PG::Connection.new(connection_string) → conn
PG::Connection.new(host, port, options, tty, dbname, user, password) → conn

Alias for: new

sync_connect(*args)

static VALUE
pgconn_s_sync_connect(int argc, VALUE *argv, VALUE klass)
{
        t_pg_connection *this;
        VALUE conninfo;
        VALUE self = pgconn_s_allocate( klass );

        this = pg_get_connection( self );
        conninfo = rb_funcall2( rb_cPGconn, rb_intern("parse_connect_args"), argc, argv );
        this->pgconn = gvl_PQconnectdb(StringValueCStr(conninfo));

        if(this->pgconn == NULL)
                rb_raise(rb_ePGerror, "PQconnectdb() unable to allocate PGconn structure");

        if (PQstatus(this->pgconn) == CONNECTION_BAD)
                pg_raise_conn_error( rb_eConnectionBad, self, "%s", PQerrorMessage(this->pgconn));

        pgconn_set_default_encoding( self );

        if (rb_block_given_p()) {
                return rb_ensure(rb_yield, self, pgconn_finish, self);
        }
        return self;
}

sync_ping(*args)

static VALUE
pgconn_s_sync_ping( int argc, VALUE *argv, VALUE klass )
{
        PGPing ping;
        VALUE conninfo;

        conninfo = rb_funcall2( klass, rb_intern("parse_connect_args"), argc, argv );
        ping     = gvl_PQping( StringValueCStr(conninfo) );

        return INT2FIX((int)ping);
}

PG::Connection.unescape_bytea( string )

Converts an escaped string representation of binary data into binary data — the reverse of escape_bytea. This is needed when retrieving bytea data in text format, but not when retrieving it in binary format.

static VALUE
pgconn_s_unescape_bytea(VALUE self, VALUE str)
{
        unsigned char *from, *to;
        size_t to_len;
        VALUE ret;

        UNUSED( self );

        Check_Type(str, T_STRING);
        from = (unsigned char*)StringValueCStr(str);

        to = PQunescapeBytea(from, &to_len);

        ret = rb_str_new((char*)to, to_len);
        PQfreemem(to);
        return ret;
}

Public Instance Methods

async_cancel()

Alias for: cancel

async_close_portal(p1)

Submits a request to close the specified portal, and waits for completion.

close_portal allows an application to trigger a close of a previously created portal. Closing a portal releases all of its associated resources on the server and allows its name to be reused. (pg does not provide any direct access to portals, but you can use this function to close a cursor created with a DECLARE CURSOR SQL command.)

portal_name can be “” or nil to reference the unnamed portal. It is fine if no portal exists with this name, in that case the operation is a no-op. On success, a PG::Result with status PGRES_COMMAND_OK is returned.

See also corresponding libpq function.

Available since PostgreSQL-17.

Alias for: close_portal

async_close_prepared(p1)

Submits a request to close the specified prepared statement, and waits for completion. close_prepared allows an application to close a previously prepared statement. Closing a statement releases all of its associated resources on the server and allows its name to be reused. It’s the same as using the DEALLOCATE SQL statement, but on a lower protocol level.

statement_name can be “” or nil to reference the unnamed statement. It is fine if no statement exists with this name, in that case the operation is a no-op. On success, a PG::Result with status PGRES_COMMAND_OK is returned.

See also corresponding libpq function.

Available since PostgreSQL-17.

Alias for: close_prepared

async_describe_portal(p1)

Retrieve information about the portal portal_name.

See also corresponding libpq function.

Alias for: describe_portal

async_describe_prepared(p1)

Retrieve information about the prepared statement statement_name.

See also corresponding libpq function.

Alias for: describe_prepared

async_encrypt_password( password, username, algorithm=nil )

Alias for: encrypt_password

async_exec(*args)

Sends SQL query request specified by sql to PostgreSQL. On success, it returns a PG::Result instance with all result rows and columns. On failure, it raises a PG::Error.

For backward compatibility, if you pass more than one parameter to this method, it will call exec_params for you. New code should explicitly use exec_params if argument placeholders are used.

If the optional code block is given, it will be passed result as an argument, and the PG::Result object will automatically be cleared when the block terminates. In this instance, conn.exec returns the value of the block.

exec is an alias for async_exec which is almost identical to sync_exec . sync_exec is implemented on the simpler synchronous command processing API of libpq, whereas async_exec is implemented on the asynchronous API and on ruby’s IO mechanisms. Only async_exec is compatible to Fiber.scheduler based asynchronous IO processing introduced in ruby-3.0. Both methods ensure that other threads can process while waiting for the server to complete the request, but sync_exec blocks all signals to be processed until the query is finished. This is most notably visible by a delayed reaction to Control+C. It’s not recommended to use explicit sync or async variants but exec instead, unless you have a good reason to do so.

See also corresponding libpq function.

Also aliased as: async_query

Alias for: exec

async_exec_params(*args)

Sends SQL query request specified by sql to PostgreSQL using placeholders for parameters.

Returns a PG::Result instance on success. On failure, it raises a PG::Error.

params is an array of the bind parameters for the SQL query. Each element of the params array may be either:

a hash of the form:
  {:value  => String (value of bind parameter)
   :type   => Integer (oid of type of bind parameter)
   :format => Integer (0 for text, 1 for binary)
  }
or, it may be a String. If it is a string, that is equivalent to the hash:
  { :value => <string value>, :type => 0, :format => 0 }

PostgreSQL bind parameters are represented as $1, $2, $3, etc., inside the SQL query. The 0th element of the params array is bound to $1, the 1st element is bound to $2, etc. nil is treated as NULL.

If the types are not specified, they will be inferred by PostgreSQL. Instead of specifying type oids, it’s recommended to simply add explicit casts in the query to ensure that the right type is used.

For example: “SELECT $1::int”

The optional result_format should be 0 for text results, 1 for binary.

type_map can be a PG::TypeMap derivation (such as PG::BasicTypeMapForQueries). This will type cast the params from various Ruby types before transmission based on the encoders defined by the type map. When a type encoder is used the format and oid of a given bind parameter are retrieved from the encoder instead out of the hash form described above.

The primary advantage of exec_params over exec is that parameter values can be separated from the command string, thus avoiding the need for tedious and error-prone quoting and escaping. Unlike exec, exec_params allows at most one SQL command in the given string. (There can be semicolons in it, but not more than one nonempty command.) This is a limitation of the underlying protocol, but has some usefulness as an extra defense against SQL-injection attacks.

See also corresponding libpq function.

Alias for: exec_params

async_exec_prepared(*args)

Execute prepared named statement specified by statement_name. Returns a PG::Result instance on success. On failure, it raises a PG::Error.

params is an array of the optional bind parameters for the SQL query. Each element of the params array may be either:

a hash of the form:
  {:value  => String (value of bind parameter)
   :format => Integer (0 for text, 1 for binary)
  }
or, it may be a String. If it is a string, that is equivalent to the hash:
  { :value => <string value>, :format => 0 }

The optional result_format should be 0 for text results, 1 for binary.

See also corresponding libpq function.

Alias for: exec_prepared

async_flush()

Attempts to flush any queued output data to the server. Returns true if data is successfully flushed, false if not. It can only return false if connection is in nonblocking mode. Raises PG::Error if some other failure occurred.

Alias for: flush

async_get_copy_data(async=false, decoder=nil)

Alias for: get_copy_data

async_get_last_result()

This function retrieves all available results on the current connection (from previously issued asynchronous commands like +send_query()+) and returns the last non-NULL result, or nil if no results are available.

If the last result contains a bad result_status, an appropriate exception is raised.

This function is similar to get_result except that it is designed to get one and only one result and that it checks the result state.

Alias for: get_last_result

async_get_result()

Alias for: get_result

async_isnonblocking()

Alias for: isnonblocking

pipeline_sync

Alias for: pipeline_sync

async_prepare(*args)

Prepares statement sql with name name to be executed later. Returns a PG::Result instance on success. On failure, it raises a PG::Error.

param_types is an optional parameter to specify the Oids of the types of the parameters.

For example: “SELECT $1::int”

PostgreSQL bind parameters are represented as $1, $2, $3, etc., inside the SQL query.

See also corresponding libpq function.

Alias for: prepare

async_put_copy_data(buffer, encoder=nil)

Alias for: put_copy_data

async_put_copy_end(*args)

Alias for: put_copy_end

async_query(*args)

Sends SQL query request specified by sql to PostgreSQL. On success, it returns a PG::Result instance with all result rows and columns. On failure, it raises a PG::Error.

For backward compatibility, if you pass more than one parameter to this method, it will call exec_params for you. New code should explicitly use exec_params if argument placeholders are used.

See also corresponding libpq function.

Alias for: async_exec

async_reset()

Alias for: reset

async_set_client_encoding(p1)

Sets the client encoding to the encoding String.

Alias for: set_client_encoding

async_setnonblocking(enabled)

Alias for: setnonblocking

backend_key() → Integer

Returns the key of the backend server process for this connection. This key can be used to cancel queries on the server.

static VALUE
pgconn_backend_key(VALUE self)
{
        int be_key;
        struct pg_cancel *cancel;
        PGconn *conn = pg_get_pgconn(self);

        cancel = (struct pg_cancel*)PQgetCancel(conn);
        if(cancel == NULL)
                pg_raise_conn_error( rb_ePGerror, self, "Invalid connection!");

        if( cancel->be_pid != PQbackendPID(conn) )
                rb_raise(rb_ePGerror,"Unexpected binary struct layout - please file a bug report at ruby-pg!");

        be_key = cancel->be_key;

        PQfreeCancel(cancel);

        return INT2NUM(be_key);
}

backend_pid() → Integer

Returns the process ID of the backend server process for this connection. Note that this is a PID on database server host.

static VALUE
pgconn_backend_pid(VALUE self)
{
        return INT2NUM(PQbackendPID(pg_get_pgconn(self)));
}

block( [ timeout ] ) → Boolean

Blocks until the server is no longer busy, or until the optional timeout is reached, whichever comes first. timeout is measured in seconds and can be fractional.

Returns false if timeout is reached, true otherwise.

If true is returned, conn.is_busy will return false and conn.get_result will not block.

VALUE
pgconn_block( int argc, VALUE *argv, VALUE self ) {
        struct timeval timeout;
        struct timeval *ptimeout = NULL;
        VALUE timeout_in;
        double timeout_sec;
        void *ret;

        if ( rb_scan_args(argc, argv, "01", &timeout_in) == 1 ) {
                timeout_sec = NUM2DBL( timeout_in );
                timeout.tv_sec = (time_t)timeout_sec;
                timeout.tv_usec = (suseconds_t)((timeout_sec - (long)timeout_sec) * 1e6);
                ptimeout = &timeout;
        }

        ret = wait_socket_readable( self, ptimeout, get_result_readable);

        if( !ret )
                return Qfalse;

        return Qtrue;
}

cancel() → String

Requests cancellation of the command currently being processed.

Returns nil on success, or a string containing the error message if a failure occurs.

On PostgreSQL-17+ client libaray the class PG::CancelConnection is used. On older client library a pure ruby implementation is used.

# File lib/pg/connection.rb, line 625
def cancel
        cancon = PG::CancelConnection.new(self)
        cancon.async_cancel
rescue PG::Error => err
        err.to_s
end

Also aliased as: async_cancel

check_socket()

Read all pending socket input to internal memory and raise an exception in case of errors.

This verifies that the connection socket is in a usable state and not aborted in any way. No communication is done with the server. Only pending data is read from the socket - the method doesn’t wait for any outstanding server answers.

Raises a kind of PG::Error if there was an error reading the data or if the socket is in a failure state.

The method doesn’t verify that the server is still responding. To verify that the communication to the server works, it is recommended to use something like conn.exec('') instead.

# File lib/pg/connection.rb, line 383
def check_socket
        while socket_io.wait_readable(0)
                consume_input
        end
        nil
end

client_encoding=(p1)

Sets the client encoding to the encoding String.

Alias for: set_client_encoding

finish

Closes the backend connection.

Alias for: finish

close_portal( portal_name ) → PG::Result

Submits a request to close the specified portal, and waits for completion.

See also corresponding libpq function.

Available since PostgreSQL-17.

static VALUE
pgconn_async_close_portal(VALUE self, VALUE portal)
{
        return pgconn_async_describe_close_prepared_potral(self, portal, pgconn_send_close_portal);
}

Also aliased as: async_close_portal

close_prepared( statement_name ) → PG::Result

See also corresponding libpq function.

Available since PostgreSQL-17.

static VALUE
pgconn_async_close_prepared(VALUE self, VALUE stmt_name)
{
        return pgconn_async_describe_close_prepared_potral(self, stmt_name, pgconn_send_close_prepared);
}

Also aliased as: async_close_prepared

conndefaults()

Returns an array of Hashes with connection defaults. See ::conndefaults for details.

# File lib/pg/connection.rb, line 328
def conndefaults
        return self.class.conndefaults
end

conndefaults_hash()

Returns a Hash with connection defaults. See ::conndefaults_hash for details.

# File lib/pg/connection.rb, line 344
def conndefaults_hash
        return self.class.conndefaults_hash
end

connect_poll() → Integer

Returns one of:

PGRES_POLLING_READING

wait until the socket is ready to read

PGRES_POLLING_WRITING

wait until the socket is ready to write

PGRES_POLLING_FAILED

the asynchronous connection has failed

PGRES_POLLING_OK

the asynchronous connection is ready

Example:

require "io/wait"

conn = PG::Connection.connect_start(dbname: 'mydatabase')
status = conn.connect_poll
while(status != PG::PGRES_POLLING_OK) do
  # do some work while waiting for the connection to complete
  if(status == PG::PGRES_POLLING_READING)
    unless conn.socket_io.wait_readable(10.0)
      raise "Asynchronous connection timed out!"
    end
  elsif(status == PG::PGRES_POLLING_WRITING)
    unless conn.socket_io.wait_writable(10.0)
      raise "Asynchronous connection timed out!"
    end
  end
  status = conn.connect_poll
end
# now conn.status == CONNECTION_OK, and connection
# is ready.

static VALUE
pgconn_connect_poll(VALUE self)
{
        PostgresPollingStatusType status;

        pgconn_close_socket_io(self);
        status = gvl_PQconnectPoll(pg_get_pgconn(self));

        return INT2FIX((int)status);
}

connection_needs_password() → Boolean

Returns true if the authentication method required a password, but none was available. false otherwise.

static VALUE
pgconn_connection_needs_password(VALUE self)
{
        return PQconnectionNeedsPassword(pg_get_pgconn(self)) ? Qtrue : Qfalse;
}

connection_used_password() → Boolean

Returns true if the authentication method used a caller-supplied password, false otherwise.

static VALUE
pgconn_connection_used_password(VALUE self)
{
        return PQconnectionUsedPassword(pg_get_pgconn(self)) ? Qtrue : Qfalse;
}

conninfo → hash

Returns the connection options used by a live connection.

static VALUE
pgconn_conninfo( VALUE self )
{
        PGconn *conn = pg_get_pgconn(self);
        PQconninfoOption *options = PQconninfo( conn );
        VALUE array = pgconn_make_conninfo_array( options );

        PQconninfoFree(options);

        return array;
}

conninfo_hash()

Return the Postgres connection info structure as a Hash keyed by option keyword (as a Symbol).

See also corresponding libpq function.

static VALUE
pgconn_async_describe_portal(VALUE self, VALUE portal)
{
        return pgconn_async_describe_close_prepared_potral(self, portal, pgconn_send_describe_portal);
}

Also aliased as: async_describe_portal

describe_prepared( statement_name ) → PG::Result

Retrieve information about the prepared statement statement_name.

See also corresponding libpq function.

static VALUE
pgconn_async_describe_prepared(VALUE self, VALUE stmt_name)
{
        return pgconn_async_describe_close_prepared_potral(self, stmt_name, pgconn_send_describe_prepared);
}

Also aliased as: async_describe_prepared

discard_results()

Silently discard any prior query result that application didn’t eat. This is internally used prior to Connection#exec and sibling methods. It doesn’t raise an exception on connection errors, but returns false instead.

Returns:

nil when the connection is already idle
true when some results have been discarded
false when a failure occurred and the connection was closed

static VALUE
pgconn_discard_results(VALUE self)
{
        PGconn *conn = pg_get_pgconn(self);
        VALUE socket_io;

        switch( PQtransactionStatus(conn) ) {
                case PQTRANS_IDLE:
                case PQTRANS_INTRANS:
                case PQTRANS_INERROR:
                        return Qnil;
                default:;
        }

        socket_io = pgconn_socket_io(self);

        for(;;) {
                PGresult *cur;
                int status;

                /* pgconn_block() raises an exception in case of errors.
                * To avoid this call pg_rb_io_wait() and PQconsumeInput() without rb_raise().
                */
                while( gvl_PQisBusy(conn) ){
                        int events;

                        switch( PQflush(conn) ) {
                                case 1:
                                        events = RB_NUM2INT(pg_rb_io_wait(socket_io, RB_INT2NUM(PG_RUBY_IO_READABLE | PG_RUBY_IO_WRITABLE), Qnil));
                                        if (events & PG_RUBY_IO_READABLE){
                                                if ( PQconsumeInput(conn) == 0 ) goto error;
                                        }
                                        break;
                                case 0:
                                        pg_rb_io_wait(socket_io, RB_INT2NUM(PG_RUBY_IO_READABLE), Qnil);
                                        if ( PQconsumeInput(conn) == 0 ) goto error;
                                        break;
                                default:
                                        goto error;
                        }
                }

                cur = gvl_PQgetResult(conn);
                if( cur == NULL) break;

                status = PQresultStatus(cur);
                PQclear(cur);
                if (status == PGRES_COPY_IN){
                        while( gvl_PQputCopyEnd(conn, "COPY terminated by new query or discard_results") == 0 ){
                                pgconn_async_flush(self);
                        }
                }
                if (status == PGRES_COPY_OUT){
                        for(;;) {
                                char *buffer = NULL;
                                int st = gvl_PQgetCopyData(conn, &buffer, 1);
                                if( st == 0 ) {
                                        /* would block -> wait for readable data */
                                        pg_rb_io_wait(socket_io, RB_INT2NUM(PG_RUBY_IO_READABLE), Qnil);
                                        if ( PQconsumeInput(conn) == 0 ) goto error;
                                } else if( st > 0 ) {
                                        /* some data retrieved -> discard it */
                                        PQfreemem(buffer);
                                } else {
                                        /* no more data */
                                        break;
                                }
                        }
                }
        }

        return Qtrue;

error:
        pgconn_close_socket_io(self);
        return Qfalse;
}

encoder_for_put_copy_data → PG::Coder

Returns the default coder object that is currently set for type casting of parameters to put_copy_data .

Returns either:

a kind of PG::Coder
nil - type encoding is disabled, data must be a String.

static VALUE
pgconn_encoder_for_put_copy_data_get(VALUE self)
{
        t_pg_connection *this = pg_get_connection( self );

        return this->encoder_for_put_copy_data;
}

encoder_for_put_copy_data = encoder

Set the default coder that is used for type casting of parameters to put_copy_data .

encoder can be:

a kind of PG::Coder
nil - disable type encoding, data must be a String.

static VALUE
pgconn_encoder_for_put_copy_data_set(VALUE self, VALUE encoder)
{
        t_pg_connection *this = pg_get_connection( self );

        rb_check_frozen(self);
        if( encoder != Qnil ){
                t_pg_coder *co;
                UNUSED(co);
                /* Check argument type */
                TypedData_Get_Struct(encoder, t_pg_coder, &pg_coder_type, co);
        }
        RB_OBJ_WRITE(self, &this->encoder_for_put_copy_data, encoder);

        return encoder;
}

encrypt_password( password, username, algorithm=nil ) → String

This function is intended to be used by client applications that wish to send commands like ALTER USER joe PASSWORD 'pwd'. It is good practice not to send the original cleartext password in such a command, because it might be exposed in command logs, activity displays, and so on. Instead, use this function to convert the password to encrypted form before it is sent.

The password and username arguments are the cleartext password, and the SQL name of the user it is for. algorithm specifies the encryption algorithm to use to encrypt the password. Currently supported algorithms are md5 and scram-sha-256 (on and off are also accepted as aliases for md5, for compatibility with older server versions). Note that support for scram-sha-256 was introduced in PostgreSQL version 10, and will not work correctly with older server versions. If algorithm is omitted or nil, this function will query the server for the current value of the password_encryption setting. That can block, and will fail if the current transaction is aborted, or if the connection is busy executing another query. If you wish to use the default algorithm for the server but want to avoid blocking, query password_encryption yourself before calling encrypt_password, and pass that value as the algorithm.

Return value is the encrypted password. The caller can assume the string doesn’t contain any special characters that would require escaping.

Available since PostgreSQL-10. See also corresponding libpq function.

# File lib/pg/connection.rb, line 578
def encrypt_password( password, username, algorithm=nil )
        algorithm ||= exec("SHOW password_encryption").getvalue(0,0)
        sync_encrypt_password(password, username, algorithm)
end

Also aliased as: async_encrypt_password

enter_pipeline_mode → nil

Causes a connection to enter pipeline mode if it is currently idle or already in pipeline mode.

Raises PG::Error and has no effect if the connection is not currently idle, i.e., it has a result ready, or it is waiting for more input from the server, etc. This function does not actually send anything to the server, it just changes the libpq connection state.

See the PostgreSQL documentation.

Available since PostgreSQL-14

static VALUE
pgconn_enter_pipeline_mode(VALUE self)
{
        PGconn *conn = pg_get_pgconn(self);
        int res = PQenterPipelineMode(conn);
        if( res != 1 )
                pg_raise_conn_error( rb_ePGerror, self, "%s", PQerrorMessage(conn));

        return Qnil;
}

error_message → String

Returns the error message most recently generated by an operation on the connection.

Nearly all libpq functions will set a message for conn.error_message if they fail. Note that by libpq convention, a nonempty error_message result can consist of multiple lines, and will include a trailing newline.

static VALUE
pgconn_error_message(VALUE self)
{
        char *error = PQerrorMessage(pg_get_pgconn(self));
        if (!error) return Qnil;
        return rb_str_new2(error);
}

escape(p1)

Returns a SQL-safe version of the String str. This is the preferred way to make strings safe for inclusion in SQL queries.

Consider using exec_params, which avoids the need for passing values inside of SQL commands.

Character encoding of escaped string will be equal to client encoding of connection.

See also convenience functions escape_literal and escape_identifier which also add proper quotes around the string.

Alias for: escape_string

escape_bytea( string ) → String

Escapes binary data for use within an SQL command with the type bytea.

Consider using exec_params, which avoids the need for passing values inside of SQL commands.

static VALUE
pgconn_s_escape_bytea(VALUE self, VALUE str)
{
        unsigned char *from, *to;
        size_t from_len, to_len;
        VALUE ret;

        Check_Type(str, T_STRING);
        from      = (unsigned char*)RSTRING_PTR(str);
        from_len  = RSTRING_LEN(str);

        if ( rb_obj_is_kind_of(self, rb_cPGconn) ) {
                to = PQescapeByteaConn(pg_get_pgconn(self), from, from_len, &to_len);
        } else {
                to = PQescapeBytea( from, from_len, &to_len);
        }

        ret = rb_str_new((char*)to, to_len - 1);
        PQfreemem(to);
        return ret;
}

Also aliased as: escape_bytea

escape_identifier( str ) → String

Escape an arbitrary String str as an identifier.

This method does the same as quote_ident with a String argument, but it doesn’t support an Array argument and it makes use of libpq to process the string.

static VALUE
pgconn_escape_identifier(VALUE self, VALUE string)
{
        t_pg_connection *this = pg_get_connection_safe( self );
        char *escaped = NULL;
        VALUE result = Qnil;
        int enc_idx = this->enc_idx;

        StringValueCStr(string);
        if( ENCODING_GET(string) != enc_idx ){
                string = rb_str_export_to_enc(string, rb_enc_from_index(enc_idx));
        }

        escaped = PQescapeIdentifier(this->pgconn, RSTRING_PTR(string), RSTRING_LEN(string));
        if (escaped == NULL)
                pg_raise_conn_error( rb_ePGerror, self, "%s", PQerrorMessage(this->pgconn));

        result = rb_str_new2(escaped);
        PQfreemem(escaped);
        PG_ENCODING_SET_NOCHECK(result, enc_idx);

        return result;
}

escape_literal( str ) → String

Escape an arbitrary String str as a literal.

See also PG::TextEncoder::QuotedLiteral for a type cast integrated version of this function.

static VALUE
pgconn_escape_literal(VALUE self, VALUE string)
{
        t_pg_connection *this = pg_get_connection_safe( self );
        char *escaped = NULL;
        VALUE result = Qnil;
        int enc_idx = this->enc_idx;

        StringValueCStr(string);
        if( ENCODING_GET(string) != enc_idx ){
                string = rb_str_export_to_enc(string, rb_enc_from_index(enc_idx));
        }

        escaped = PQescapeLiteral(this->pgconn, RSTRING_PTR(string), RSTRING_LEN(string));
        if (escaped == NULL)
                pg_raise_conn_error( rb_ePGerror, self, "%s", PQerrorMessage(this->pgconn));

        result = rb_str_new2(escaped);
        PQfreemem(escaped);
        PG_ENCODING_SET_NOCHECK(result, enc_idx);

        return result;
}

escape_string( str ) → String

Returns a SQL-safe version of the String str. This is the preferred way to make strings safe for inclusion in SQL queries.

Consider using exec_params, which avoids the need for passing values inside of SQL commands.

Character encoding of escaped string will be equal to client encoding of connection.

See also convenience functions escape_literal and escape_identifier which also add proper quotes around the string.

static VALUE
pgconn_s_escape(VALUE self, VALUE string)
{
        size_t size;
        int error;
        VALUE result;
        int enc_idx;
        int singleton = !rb_obj_is_kind_of(self, rb_cPGconn);

        StringValueCStr(string);
        enc_idx = singleton ? ENCODING_GET(string) : pg_get_connection(self)->enc_idx;
        if( ENCODING_GET(string) != enc_idx ){
                string = rb_str_export_to_enc(string, rb_enc_from_index(enc_idx));
        }

        result = rb_str_new(NULL, RSTRING_LEN(string) * 2 + 1);
        PG_ENCODING_SET_NOCHECK(result, enc_idx);
        if( !singleton ) {
                size = PQescapeStringConn(pg_get_pgconn(self), RSTRING_PTR(result),
                        RSTRING_PTR(string), RSTRING_LEN(string), &error);
                if(error)
                        pg_raise_conn_error( rb_ePGerror, self, "%s", PQerrorMessage(pg_get_pgconn(self)));

        } else {
                size = PQescapeString(RSTRING_PTR(result), RSTRING_PTR(string), RSTRING_LEN(string));
        }
        rb_str_set_len(result, size);

        return result;
}

Also aliased as: escape_string , escape

exec(sql) → PG::Result
exec(sql) {|pg_result| block }

Sends SQL query request specified by sql to PostgreSQL. On success, it returns a PG::Result instance with all result rows and columns. On failure, it raises a PG::Error.

For backward compatibility, if you pass more than one parameter to this method, it will call exec_params for you. New code should explicitly use exec_params if argument placeholders are used.

See also corresponding libpq function.

static VALUE
pgconn_async_exec(int argc, VALUE *argv, VALUE self)
{
        VALUE rb_pgresult = Qnil;

        pgconn_discard_results( self );
        pgconn_send_query( argc, argv, self );
        rb_pgresult = pgconn_async_get_last_result( self );

        if ( rb_block_given_p() ) {
                return rb_ensure( rb_yield, rb_pgresult, pg_result_clear, rb_pgresult );
        }
        return rb_pgresult;
}

Also aliased as: async_exec

exec_params(sql, params [, result_format [, type_map ]] ) → nil
exec_params(sql, params [, result_format [, type_map ]] ) {|pg_result| block }

Sends SQL query request specified by sql to PostgreSQL using placeholders for parameters.

Returns a PG::Result instance on success. On failure, it raises a PG::Error.

params is an array of the bind parameters for the SQL query. Each element of the params array may be either:

a hash of the form:
  {:value  => String (value of bind parameter)
   :type   => Integer (oid of type of bind parameter)
   :format => Integer (0 for text, 1 for binary)
  }
or, it may be a String. If it is a string, that is equivalent to the hash:
  { :value => <string value>, :type => 0, :format => 0 }

For example: “SELECT $1::int”

The optional result_format should be 0 for text results, 1 for binary.

See also corresponding libpq function.

static VALUE
pgconn_async_exec_params(int argc, VALUE *argv, VALUE self)
{
        VALUE rb_pgresult = Qnil;

        pgconn_discard_results( self );
        /* If called with no or nil parameters, use PQsendQuery for compatibility */
        if ( argc == 1 || (argc >= 2 && argc <= 4 && NIL_P(argv[1]) )) {
                pg_deprecated(3, ("forwarding async_exec_params to async_exec is deprecated"));
                pgconn_send_query( argc, argv, self );
        } else {
                pgconn_send_query_params( argc, argv, self );
        }
        rb_pgresult = pgconn_async_get_last_result( self );

        if ( rb_block_given_p() ) {
                return rb_ensure( rb_yield, rb_pgresult, pg_result_clear, rb_pgresult );
        }
        return rb_pgresult;
}

Also aliased as: async_exec_params

exec_prepared(statement_name [, params, result_format[, type_map]] ) → PG::Result
exec_prepared(statement_name [, params, result_format[, type_map]] ) {|pg_result| block }

Execute prepared named statement specified by statement_name. Returns a PG::Result instance on success. On failure, it raises a PG::Error.

params is an array of the optional bind parameters for the SQL query. Each element of the params array may be either:

a hash of the form:
  {:value  => String (value of bind parameter)
   :format => Integer (0 for text, 1 for binary)
  }
or, it may be a String. If it is a string, that is equivalent to the hash:
  { :value => <string value>, :format => 0 }

The optional result_format should be 0 for text results, 1 for binary.

See also corresponding libpq function.

static VALUE
pgconn_async_exec_prepared(int argc, VALUE *argv, VALUE self)
{
        VALUE rb_pgresult = Qnil;

        pgconn_discard_results( self );
        pgconn_send_query_prepared( argc, argv, self );
        rb_pgresult = pgconn_async_get_last_result( self );

        if ( rb_block_given_p() ) {
                return rb_ensure( rb_yield, rb_pgresult, pg_result_clear, rb_pgresult );
        }
        return rb_pgresult;
}

Also aliased as: async_exec_prepared

exit_pipeline_mode → nil

Causes a connection to exit pipeline mode if it is currently in pipeline mode with an empty queue and no pending results.

Takes no action if not in pipeline mode. Raises PG::Error if the current statement isn’t finished processing, or PQgetResult has not been called to collect results from all previously sent query.

Available since PostgreSQL-14

static VALUE
pgconn_exit_pipeline_mode(VALUE self)
{
        PGconn *conn = pg_get_pgconn(self);
        int res = PQexitPipelineMode(conn);
        if( res != 1 )
                pg_raise_conn_error( rb_ePGerror, self, "%s", PQerrorMessage(conn));

        return Qnil;
}

external_encoding() → Encoding

Return the server_encoding of the connected database as a Ruby Encoding object. The SQL_ASCII encoding is mapped to to ASCII_8BIT.

static VALUE
pgconn_external_encoding(VALUE self)
{
        t_pg_connection *this = pg_get_connection_safe( self );
        rb_encoding *enc = NULL;
        const char *pg_encname = NULL;

        pg_encname = PQparameterStatus( this->pgconn, "server_encoding" );
        enc = pg_get_pg_encname_as_rb_encoding( pg_encname );
        return rb_enc_from_encoding( enc );
}

field_name_type → Symbol

Get type of field names.

See description at field_name_type=

static VALUE
pgconn_field_name_type_get(VALUE self)
{
        t_pg_connection *this = pg_get_connection( self );

        if( this->flags & PG_RESULT_FIELD_NAMES_SYMBOL ){
                return sym_symbol;
        } else if( this->flags & PG_RESULT_FIELD_NAMES_STATIC_SYMBOL ){
                return sym_static_symbol;
        } else {
                return sym_string;
        }
}

field_name_type = Symbol

Set default type of field names of results retrieved by this connection. It can be set to one of:

:string to use String based field names
:symbol to use Symbol based field names

The default is :string .

Settings the type of field names affects only future results.

See further description at PG::Result#field_name_type=

static VALUE
pgconn_field_name_type_set(VALUE self, VALUE sym)
{
        t_pg_connection *this = pg_get_connection( self );

        rb_check_frozen(self);
        this->flags &= ~PG_RESULT_FIELD_NAMES_MASK;
        if( sym == sym_symbol ) this->flags |= PG_RESULT_FIELD_NAMES_SYMBOL;
        else if ( sym == sym_static_symbol ) this->flags |= PG_RESULT_FIELD_NAMES_STATIC_SYMBOL;
        else if ( sym == sym_string );
        else rb_raise(rb_eArgError, "invalid argument %+"PRIsVALUE, sym);

        return sym;
}

finish

Closes the backend connection.

static VALUE
pgconn_finish( VALUE self )
{
        t_pg_connection *this = pg_get_connection_safe( self );

        pgconn_close_socket_io( self );
        PQfinish( this->pgconn );
        this->pgconn = NULL;
        return Qnil;
}

Also aliased as: close

finished? → boolean

Returns true if the backend connection has been closed.

static VALUE
pgconn_finished_p( VALUE self )
{
        t_pg_connection *this = pg_get_connection( self );
        if ( this->pgconn ) return Qfalse;
        return Qtrue;
}

flush() → Boolean

static VALUE
pgconn_async_flush(VALUE self)
{
        while( pgconn_sync_flush(self) == Qfalse ){
                /* wait for the socket to become read- or write-ready */
                int events;
                VALUE socket_io = pgconn_socket_io(self);
                events = RB_NUM2INT(pg_rb_io_wait(socket_io, RB_INT2NUM(PG_RUBY_IO_READABLE | PG_RUBY_IO_WRITABLE), Qnil));

                if (events & PG_RUBY_IO_READABLE){
                        pgconn_consume_input(self);
                }
        }
        return Qtrue;
}

Also aliased as: async_flush

get_client_encoding() → String

Returns the client encoding as a String.

static VALUE
pgconn_get_client_encoding(VALUE self)
{
        char *encoding = (char *)pg_encoding_to_char(PQclientEncoding(pg_get_pgconn(self)));
        return rb_str_new2(encoding);
}

get_copy_data( [ nonblock = false [, decoder = nil ]] ) → Object

Return one row of data, nil if the copy is done, or false if the call would block (only possible if nonblock is true).

If decoder is not set or nil, data is returned as binary string.

If decoder is set to a PG::Coder derivation, the return type depends on this decoder. PG::TextDecoder::CopyRow decodes the received data fields from one row of PostgreSQL’s COPY text format to an Array of Strings. Optionally the decoder can type cast the single fields to various Ruby types in one step, if PG::TextDecoder::CopyRow#type_map is set accordingly.

See also corresponding libpq function.

static VALUE
pgconn_async_prepare(int argc, VALUE *argv, VALUE self)
{
        VALUE rb_pgresult = Qnil;

        pgconn_discard_results( self );
        pgconn_send_prepare( argc, argv, self );
        rb_pgresult = pgconn_async_get_last_result( self );

        if ( rb_block_given_p() ) {
                return rb_ensure( rb_yield, rb_pgresult, pg_result_clear, rb_pgresult );
        }
        return rb_pgresult;
}

Also aliased as: async_prepare

protocol_version → Integer

Interrogates the frontend/backend protocol being used.

Applications might wish to use this function to determine whether certain features are supported. Currently, the only value is 3 (3.0 protocol). The protocol version will not change after connection startup is complete, but it could theoretically change during a connection reset. The 3.0 protocol is supported by PostgreSQL server versions 7.4 and above.

PG::ConnectionBad is raised if the connection is bad.

static VALUE
pgconn_protocol_version(VALUE self)
{
        int protocol_version = PQprotocolVersion(pg_get_pgconn(self));
        if (protocol_version == 0) {
                pg_raise_conn_error( rb_eConnectionBad, self, "PQprotocolVersion() can't get protocol version");
        }
        return INT2NUM(protocol_version);
}

put_copy_data( buffer [, encoder] ) → Boolean

Transmits buffer as copy data to the server. Returns true if the data was sent, false if it was not sent (false is only possible if the connection is in nonblocking mode, and this command would block).

encoder can be a PG::Coder derivation (typically PG::TextEncoder::CopyRow). This encodes the data fields given as buffer from an Array of Strings to PostgreSQL’s COPY text format inclusive proper escaping. Optionally the encoder can type cast the fields from various Ruby types in one step, if PG::TextEncoder::CopyRow#type_map is set accordingly.

Raises an exception if an error occurs.

See also corresponding libpq function.

static VALUE
pgconn_set_error_context_visibility(VALUE self, VALUE in_context_visibility)
{
        PGconn *conn = pg_get_pgconn(self);
        PGContextVisibility context_visibility = NUM2INT(in_context_visibility);
        return INT2FIX(PQsetErrorContextVisibility(conn, context_visibility));
}

set_error_verbosity( verbosity ) → Integer

Sets connection’s verbosity to verbosity and returns the previous setting. Available settings are:

PQERRORS_TERSE
PQERRORS_DEFAULT
PQERRORS_VERBOSE
PQERRORS_SQLSTATE

Changing the verbosity does not affect the messages available from already-existing PG::Result objects, only subsequently-created ones. (But see PG::Result#verbose_error_message if you want to print a previous error with a different verbosity.)

See also corresponding libpq function.

static VALUE
pgconn_set_error_verbosity(VALUE self, VALUE in_verbosity)
{
        PGconn *conn = pg_get_pgconn(self);
        PGVerbosity verbosity = NUM2INT(in_verbosity);
        return INT2FIX(PQsetErrorVerbosity(conn, verbosity));
}

set_notice_processor {|message| ... } → Proc

See set_notice_receiver for the description of what this and the notice_processor methods do.

This function takes a new block to act as the notice processor and returns the Proc object previously set, or nil if it was previously the default. The block should accept a single String object.

If you pass no arguments, it will reset the handler to the default.

static VALUE
pgconn_set_notice_processor(VALUE self)
{
        VALUE proc, old_proc;
        t_pg_connection *this = pg_get_connection_safe( self );

        rb_check_frozen(self);
        /* If default_notice_processor is unset, assume that the current
         * notice processor is the default, and save it to a global variable.
         * This should not be a problem because the default processor is
         * always the same, so won't vary among connections.
         */
        if(this->default_notice_processor == NULL)
                this->default_notice_processor = PQsetNoticeProcessor(this->pgconn, NULL, NULL);

        old_proc = this->notice_processor;
        if( rb_block_given_p() ) {
                proc = rb_block_proc();
                PQsetNoticeProcessor(this->pgconn, gvl_notice_processor_proxy, (void *)self);
        } else {
                /* if no block is given, set back to default */
                proc = Qnil;
                PQsetNoticeProcessor(this->pgconn, this->default_notice_processor, NULL);
        }

        RB_OBJ_WRITE(self, &this->notice_processor, proc);
        return old_proc;
}

set_notice_receiver {|result| ... } → Proc

Notice and warning messages generated by the server are not returned by the query execution functions, since they do not imply failure of the query. Instead they are passed to a notice handling function, and execution continues normally after the handler returns. The default notice handling function prints the message on stderr, but the application can override this behavior by supplying its own handling function.

For historical reasons, there are two levels of notice handling, called the notice receiver and notice processor. The default behavior is for the notice receiver to format the notice and pass a string to the notice processor for printing. However, an application that chooses to provide its own notice receiver will typically ignore the notice processor layer and just do all the work in the notice receiver.

This function takes a new block to act as the handler, which should accept a single parameter that will be a PG::Result object, and returns the Proc object previously set, or nil if it was previously the default.

If you pass no arguments, it will reset the handler to the default.

Note: The result passed to the block should not be used outside of the block, since the corresponding C object could be freed after the block finishes.

static VALUE
pgconn_set_notice_receiver(VALUE self)
{
        VALUE proc, old_proc;
        t_pg_connection *this = pg_get_connection_safe( self );

        rb_check_frozen(self);
        /* If default_notice_receiver is unset, assume that the current
         * notice receiver is the default, and save it to a global variable.
         * This should not be a problem because the default receiver is
         * always the same, so won't vary among connections.
         */
        if(this->default_notice_receiver == NULL)
                this->default_notice_receiver = PQsetNoticeReceiver(this->pgconn, NULL, NULL);

        old_proc = this->notice_receiver;
        if( rb_block_given_p() ) {
                proc = rb_block_proc();
                PQsetNoticeReceiver(this->pgconn, gvl_notice_receiver_proxy, (void *)self);
        } else {
                /* if no block is given, set back to default */
                proc = Qnil;
                PQsetNoticeReceiver(this->pgconn, this->default_notice_receiver, NULL);
        }

        RB_OBJ_WRITE(self, &this->notice_receiver, proc);
        return old_proc;
}

set_single_row_mode → self

To enter single-row mode, call this method immediately after a successful call of send_query (or a sibling function). This mode selection is effective only for the currently executing query. Then call Connection#get_result repeatedly, until it returns nil.

Each (but the last) received Result has exactly one row and a Result#result_status of PGRES_SINGLE_TUPLE. The last Result has zero rows and is used to indicate a successful execution of the query. All of these Result objects will contain the same row description data (column names, types, etc) that an ordinary Result object for the query would have.

Caution: While processing a query, the server may return some rows and then encounter an error, causing the query to be aborted. Ordinarily, pg discards any such rows and reports only the error. But in single-row or chunked mode, some rows may have already been returned to the application. Hence, the application will see some PGRES_SINGLE_TUPLE or PGRES_TUPLES_CHUNK PG::Result objects followed by a PG::Error raised in get_result. For proper transactional behavior, the application must be designed to discard or undo whatever has been done with the previously-processed rows, if the query ultimately fails.

Example:

conn.send_query( "your SQL command" )
conn.set_single_row_mode
loop do
  res = conn.get_result or break
  res.check
  res.each do |row|
    # do something with the received row
  end
end

static VALUE
pgconn_set_single_row_mode(VALUE self)
{
        PGconn *conn = pg_get_pgconn(self);

        rb_check_frozen(self);
        if( PQsetSingleRowMode(conn) == 0 )
                pg_raise_conn_error( rb_ePGerror, self, "PQsetSingleRowMode %s", PQerrorMessage(conn));

        return self;
}

setnonblocking(Boolean) → nil

Sets the nonblocking status of the connection. In the blocking state, calls to send_query will block until the message is sent to the server, but will not wait for the query results. In the nonblocking state, calls to send_query will return an error if the socket is not ready for writing. Note: This function does not affect exec, because that function doesn’t return until the server has processed the query and returned the results.

Returns nil.

# File lib/pg/connection.rb, line 461
def setnonblocking(enabled)
        singleton_class.async_send_api = !enabled
        self.flush_data = !enabled
        sync_setnonblocking(true)
end

Also aliased as: async_setnonblocking

socket() → Integer

This method is deprecated. Please use the more portable method socket_io .

Returns the socket’s file descriptor for this connection. IO.for_fd() can be used to build a proper IO object to the socket. If you do so, you will likely also want to set autoclose=false on it to prevent Ruby from closing the socket to PostgreSQL if it goes out of scope. Alternatively, you can use socket_io, which creates an IO that’s associated with the connection object itself, and so won’t go out of scope until the connection does.

Note: On Windows the file descriptor is not usable, since it can not be used to build a Ruby IO object.

static VALUE
pgconn_socket(VALUE self)
{
        int sd;
        pg_deprecated(4, ("conn.socket is deprecated and should be replaced by conn.socket_io"));

        if( (sd = PQsocket(pg_get_pgconn(self))) < 0)
                pg_raise_conn_error( rb_eConnectionBad, self, "PQsocket() can't get socket descriptor");

        return INT2NUM(sd);
}

socket_io() → IO

Fetch an IO object created from the Connection’s underlying socket. This object can be used per socket_io.wait_readable, socket_io.wait_writable or for IO.select to wait for events while running asynchronous API calls. IO#wait_*able is Fiber.scheduler compatible in contrast to IO.select.

The IO object can change while the connection is established, but is memorized afterwards. So be sure not to cache the IO object, but repeat calling conn.socket_io instead.

Using this method also works on Windows in contrast to using socket . It also avoids the problem of the underlying connection being closed by Ruby when an IO created using IO.for_fd(conn.socket) goes out of scope.

static VALUE
pgconn_socket_io(VALUE self)
{
        t_pg_connection *this = pg_get_connection_safe( self );

        if ( !RTEST(this->socket_io) ) {
                int sd;
                if( (sd = PQsocket(this->pgconn)) < 0){
                        pg_raise_conn_error( rb_eConnectionBad, self, "PQsocket() can't get socket descriptor");
                }
                return pg_wrap_socket_io( sd, self, &this->socket_io, &this->ruby_sd);
        }

        return this->socket_io;
}

ssl_attribute(attribute_name) → String

Returns SSL-related information about the connection.

The list of available attributes varies depending on the SSL library being used, and the type of connection. If an attribute is not available, returns nil.

The following attributes are commonly available:

library

Name of the SSL implementation in use. (Currently, only “OpenSSL” is implemented)

protocol

SSL/TLS version in use. Common values are “SSLv2”, “SSLv3”, “TLSv1”, “TLSv1.1” and “TLSv1.2”, but an implementation may return other strings if some other protocol is used.

key_bits

Number of key bits used by the encryption algorithm.

cipher

A short name of the ciphersuite used, e.g. “DHE-RSA-DES-CBC3-SHA”. The names are specific to each SSL implementation.

compression

If SSL compression is in use, returns the name of the compression algorithm, or “on” if compression is used but the algorithm is not known. If compression is not in use, returns “off”.

static VALUE
pgconn_ssl_attribute(VALUE self, VALUE attribute_name)
{
        const char *p_attr;

        p_attr = PQsslAttribute(pg_get_pgconn(self), StringValueCStr(attribute_name));
        return p_attr ? rb_str_new_cstr(p_attr) : Qnil;
}

ssl_attribute_names → Array

Return an array of SSL attribute names available.