AcePerl
view release on metacpan or search on metacpan
Ace/Local.pm view on Meta::CPAN
sub DESTROY {
my $self = shift;
return unless kill 0,$self->{'pid'};
if ($self->auto_save) {
# save work for the user...
$self->query('save');
$self->synch;
}
$self->query('quit');
# just for paranoid reasons. shouldn't be necessary
close $self->{'write'} if $self->{'write'};
close $self->{'read'} if $self->{'read'};
waitpid($self->{pid},0) if $self->{'pid'};
}
sub encore {
my $self = shift;
return $self->status == STATUS_PENDING;
}
sub auto_save {
my $self = shift;
$self->{'auto_save'} = $_[0] if defined $_[0];
return $self->{'auto_save'};
}
sub status {
return $_[0]->{'status'};
}
sub error {
my $self = shift;
return $self->{'error'};
}
sub query {
my $self = shift;
my $query = shift;
warn "query($query)\n" if $self->debug;
if ($self->debug) {
my $msg = $query || '';
warn "\tquery($msg)";
}
return undef if $self->{'status'} == STATUS_ERROR;
do $self->read() until $self->{'status'} != STATUS_PENDING;
my $wtr = $self->{'write'};
print $wtr "$query\n";
$self->{'status'} = STATUS_PENDING;
}
sub low_read { # hack to accomodate "uninitialized database" warning from tace
my $self = shift;
my $rdr = $self->{'read'};
return undef unless $self->{'status'} == STATUS_PENDING;
my $rin = '';
my $data = '';
vec($rin,fileno($rdr),1)=1;
unless (select($rin,undef,undef,1)) {
$self->{'status'} = STATUS_WAITING;
return undef;
}
sysread($rdr,$data,READSIZE);
return $data;
}
sub read {
my $self = shift;
return undef unless $self->{'status'} == STATUS_PENDING;
my $rdr = $self->{'read'};
my $len = defined $self->{'buffer'} ? length($self->{'buffer'}) : 0;
my $plen = length($self->{'prompt'});
my ($result, $bytes, $pos, $searchfrom);
while (1) {
# Read the data directly onto the end of the buffer
$bytes = sysread($rdr, $self->{'buffer'},
READSIZE, $len);
unless ($bytes > 0) {
$self->{'status'} = STATUS_ERROR;
return;
}
# check for prompt
# The following checks were implemented using regexps and $' and
# friends. I have changed this to use {r}index and substr (a)
# because they're much faster than regexps and (b) because using
# $' and $` causes all regexps in a program to execute
# very slowly due to excessive and unnecessary pre/post-match
# copying -- tim.cutts@incyte.com 08 Sep 1999
# Note, don't need to search the whole buffer for the prompt;
# just need to search the new data and the prompt length from
# any previous data.
$searchfrom = ($len <= $plen) ? 0 : ($len - $plen);
if (($pos = index($self->{'buffer'},
$self->{'prompt'},
$searchfrom)) > 0) {
$self->{'status'} = STATUS_WAITING;
$result = substr($self->{'buffer'}, 0, $pos);
$self->{'buffer'} = '';
return $result;
}
# return partial results for paragraph breaks
if (($pos = rindex($self->{'buffer'}, "\n\n")) > 0) {
$result = substr($self->{'buffer'}, 0, $pos + 2);
$self->{'buffer'} = substr($self->{'buffer'},
$pos + 2);
return $result;
}
( run in 1.513 second using v1.01-cache-2.11-cpan-39bf76dae61 )