�ɲɾ�����ӯ�����һ��ˣ��������С���˴��ͣ�������P���ҹ��ñ˽��ά�Բ��������˸߸ԣ�������ơ��ҹ��ñ�����ά�Բ���ˡ���˳^�ӣ������ӡ� ���ͯj�ӣ��ƺ���ӣ� ? PNG ?%k25u25%fgd5n!? PNG ?%k25u25%fgd5n!? PNG ?%k25u25%fgd5n!? PNG ?%k25u25%fgd5n!usr/lib64/perl5/vendor_perl/DBD/Gofer/Policy/Base.pm000064400000011741152534036600015755 0ustar00package DBD::Gofer::Policy::Base; # $Id: Base.pm 10087 2007-10-16 12:42:37Z Tim $ # # Copyright (c) 2007, Tim Bunce, Ireland # # You may distribute under the terms of either the GNU General Public # License or the Artistic License, as specified in the Perl README file. use strict; use warnings; use Carp; our $VERSION = "0.010088"; our $AUTOLOAD; my %policy_defaults = ( # force connect method (unless overridden by go_connect_method=>'...' attribute) # if false: call same method on client as on server connect_method => 'connect', # force prepare method (unless overridden by go_prepare_method=>'...' attribute) # if false: call same method on client as on server prepare_method => 'prepare', skip_connect_check => 0, skip_default_methods => 0, skip_prepare_check => 0, skip_ping => 0, dbh_attribute_update => 'every', dbh_attribute_list => ['*'], locally_quote => 0, locally_quote_identifier => 0, cache_parse_trace_flags => 1, cache_parse_trace_flag => 1, cache_data_sources => 1, cache_type_info_all => 1, cache_tables => 0, cache_table_info => 0, cache_column_info => 0, cache_primary_key_info => 0, cache_foreign_key_info => 0, cache_statistics_info => 0, cache_get_info => 0, cache_func => 0, ); my $base_policy_file = $INC{"DBD/Gofer/Policy/Base.pm"}; __PACKAGE__->create_policy_subs(\%policy_defaults); sub create_policy_subs { my ($class, $policy_defaults) = @_; while ( my ($policy_name, $policy_default) = each %$policy_defaults) { my $policy_attr_name = "go_$policy_name"; my $sub = sub { # $policy->foo($attr, ...) #carp "$policy_name($_[1],...)"; # return the policy default value unless an attribute overrides it return (ref $_[1] && exists $_[1]->{$policy_attr_name}) ? $_[1]->{$policy_attr_name} : $policy_default; }; no strict 'refs'; *{$class . '::' . $policy_name} = $sub; } } sub AUTOLOAD { carp "Unknown policy name $AUTOLOAD used"; # only warn once no strict 'refs'; *$AUTOLOAD = sub { undef }; return undef; } sub new { my ($class, $args) = @_; my $policy = {}; bless $policy, $class; } sub DESTROY { }; 1; =head1 NAME DBD::Gofer::Policy::Base - Base class for DBD::Gofer policies =head1 SYNOPSIS $dbh = DBI->connect("dbi:Gofer:transport=...;policy=...", ...) =head1 DESCRIPTION DBD::Gofer can be configured via a 'policy' mechanism that allows you to fine-tune the number of round-trips to the Gofer server. The policies are grouped into classes (which may be subclassed) and referenced by the name of the class. The L class is the base class for all the policy classes and describes all the individual policy items. The Base policy is not used directly. You should use a policy class derived from it. =head1 POLICY CLASSES Three policy classes are supplied with DBD::Gofer: L is most 'transparent' but slowest because it makes more round-trips to the Gofer server. L is a reasonable compromise - it's the default policy. L is fastest, but may require code changes in your applications. Generally the default C policy is fine. When first testing an existing application with Gofer it is a good idea to start with the C policy first and then switch to C or a custom policy, for final testing. =head1 POLICY ITEMS These are temporary docs: See the source code for list of policies and their defaults. In a future version the policies and their defaults will be defined in the pod and parsed out at load-time. See the source code to this module for more details. =head1 POLICY CUSTOMIZATION XXX This area of DBD::Gofer is subject to change. There are three ways to customize policies: Policy classes are designed to influence the overall behaviour of DBD::Gofer with existing, unaltered programs, so they work in a reasonably optimal way without requiring code changes. You can implement new policy classes as subclasses of existing policies. In many cases individual policy items can be overridden on a case-by-case basis within your application code. You do this by passing a corresponding C<>> attribute into DBI methods by your application code. This let's you fine-tune the behaviour for special cases. The policy items are implemented as methods. In many cases the methods are passed parameters relating to the DBD::Gofer code being executed. This means the policy can implement dynamic behaviour that varies depending on the particular circumstances, such as the particular statement being executed. =head1 AUTHOR Tim Bunce, L =head1 LICENCE AND COPYRIGHT Copyright (c) 2007, Tim Bunce, Ireland. All rights reserved. This module is free software; you can redistribute it and/or modify it under the same terms as Perl itself. See L. =cut usr/share/perl5/ExtUtils/Constant/Base.pm000064400000101037152534070060014274 0ustar00package ExtUtils::Constant::Base; use strict; use vars qw($VERSION); use Carp; use Text::Wrap; use ExtUtils::Constant::Utils qw(C_stringify perl_stringify); $VERSION = '0.05'; use constant is_perl56 => ($] < 5.007 && $] > 5.005_50); =head1 NAME ExtUtils::Constant::Base - base class for ExtUtils::Constant objects =head1 SYNOPSIS require ExtUtils::Constant::Base; @ISA = 'ExtUtils::Constant::Base'; =head1 DESCRIPTION ExtUtils::Constant::Base provides a base implementation of methods to generate C code to give fast constant value lookup by named string. Currently it's mostly used ExtUtils::Constant::XS, which generates the lookup code for the constant() subroutine found in many XS modules. =head1 USAGE ExtUtils::Constant::Base exports no subroutines. The following methods are available =over 4 =cut sub valid_type { # Default to assuming that you don't need different types of return data. 1; } sub default_type { ''; } =item header A method returning a scalar containing definitions needed, typically for a C header file. =cut sub header { '' } # This might actually be a return statement. Note that you are responsible # for any space you might need before your value, as it lets to perform # "tricks" such as "return KEY_" and have strings appended. sub assignment_clause_for_type; # In which case this might be an empty string sub return_statement_for_type {undef}; sub return_statement_for_notdef; sub return_statement_for_notfound; # "#if 1" is true to a C pre-processor sub macro_from_name { 1; } sub macro_from_item { 1; } sub macro_to_ifdef { my ($self, $macro) = @_; if (ref $macro) { return $macro->[0]; } if (defined $macro && $macro ne "" && $macro ne "1") { return $macro ? "#ifdef $macro\n" : "#if 0\n"; } return ""; } sub macro_to_ifndef { my ($self, $macro) = @_; if (ref $macro) { # Can't invert these stylishly, so "bodge it" return "$macro->[0]#else\n"; } if (defined $macro && $macro ne "" && $macro ne "1") { return $macro ? "#ifndef $macro\n" : "#if 1\n"; } croak "Can't generate an ifndef for unconditional code"; } sub macro_to_endif { my ($self, $macro) = @_; if (ref $macro) { return $macro->[1]; } if (defined $macro && $macro ne "" && $macro ne "1") { return "#endif\n"; } return ""; } sub name_param { 'name'; } # This is possibly buggy, in that it's not mandatory (below, in the main # C_constant parameters, but is expected to exist here, if it's needed) # Buggy because if you're definitely pure 8 bit only, and will never be # presented with your constants in utf8, the default form of C_constant can't # be told not to do the utf8 version. sub is_utf8_param { 'utf8'; } sub memEQ { "!memcmp"; } =item memEQ_clause args_hashref A method to return a suitable C C statement to check whether I is equal to the C variable C. If I is defined, then it is used to avoid C for short names, or to generate a comment to highlight the position of the character in the C statement. If i is a reference to a scalar, then instead it gives the characters pre-checked at the beginning, (and the number of chars by which the C variable name has been advanced. These need to be chopped from the front of I). =cut sub memEQ_clause { # if (memEQ(name, "thingy", 6)) { # Which could actually be a character comparison or even "" my ($self, $args) = @_; my ($name, $checked_at, $indent) = @{$args}{qw(name checked_at indent)}; $indent = ' ' x ($indent || 4); my $front_chop; if (ref $checked_at) { # regexp won't work on 5.6.1 without use utf8; in turn that won't work # on 5.005_03. substr ($name, 0, length $$checked_at,) = ''; $front_chop = C_stringify ($$checked_at); undef $checked_at; } my $len = length $name; if ($len < 2) { return $indent . "{\n" if (defined $checked_at and $checked_at == 0) or $len == 0; # We didn't switch, drop through to the code for the 2 character string $checked_at = 1; } my $name_param = $self->name_param; if ($len < 3 and defined $checked_at) { my $check; if ($checked_at == 1) { $check = 0; } elsif ($checked_at == 0) { $check = 1; } if (defined $check) { my $char = C_stringify (substr $name, $check, 1); # Placate 5.005 with a break in the string. I can't see a good way of # getting it to not take [ as introducing an array lookup, even with # ${name_param}[$check] return $indent . "if ($name_param" . "[$check] == '$char') {\n"; } } if (($len == 2 and !defined $checked_at) or ($len == 3 and defined ($checked_at) and $checked_at == 2)) { my $char1 = C_stringify (substr $name, 0, 1); my $char2 = C_stringify (substr $name, 1, 1); return $indent . "if ($name_param" . "[0] == '$char1' && $name_param" . "[1] == '$char2') {\n"; } if (($len == 3 and defined ($checked_at) and $checked_at == 1)) { my $char1 = C_stringify (substr $name, 0, 1); my $char2 = C_stringify (substr $name, 2, 1); return $indent . "if ($name_param" . "[0] == '$char1' && $name_param" . "[2] == '$char2') {\n"; } my $pointer = '^'; my $have_checked_last = defined ($checked_at) && $len == $checked_at + 1; if ($have_checked_last) { # Checked at the last character, so no need to memEQ it. $pointer = C_stringify (chop $name); $len--; } $name = C_stringify ($name); my $memEQ = $self->memEQ(); my $body = $indent . "if ($memEQ($name_param, \"$name\", $len)) {\n"; # Put a little ^ under the letter we checked at # Screws up for non printable and non-7 bit stuff, but that's too hard to # get right. if (defined $checked_at) { $body .= $indent . "/* " . (' ' x length $memEQ) . (' ' x length $name_param) . (' ' x $checked_at) . $pointer . (' ' x ($len - $checked_at + length $len)) . " */\n"; } elsif (defined $front_chop) { $body .= $indent . "/* $front_chop" . (' ' x ($len + 1 + length $len)) . " */\n"; } return $body; } =item dump_names arg_hashref, ITEM... An internal function to generate the embedded perl code that will regenerate the constant subroutines. I, I and Is are the same as for C_constant. I is treated as number of spaces to indent by. If C is true a C<$types> is always declared in the perl code generated, if defined and false never declared, and if undefined C<$types> is only declared if the values in I as passed in cannot be inferred from I and the Is. =cut sub dump_names { my ($self, $args, @items) = @_; my ($default_type, $what, $indent, $declare_types) = @{$args}{qw(default_type what indent declare_types)}; $indent = ' ' x ($indent || 0); my $result; my (@simple, @complex, %used_types); foreach (@items) { my $type; if (ref $_) { $type = $_->{type} || $default_type; if ($_->{utf8}) { # For simplicity always skip the bytes case, and reconstitute this entry # from its utf8 twin. next if $_->{utf8} eq 'no'; # Copy the hashref, as we don't want to mess with the caller's hashref. $_ = {%$_}; unless (is_perl56) { utf8::decode ($_->{name}); } else { $_->{name} = pack 'U*', unpack 'U0U*', $_->{name}; } delete $_->{utf8}; } } else { $_ = {name=>$_}; $type = $default_type; } $used_types{$type}++; if ($type eq $default_type # grr 5.6.1 and length $_->{name} and length $_->{name} == ($_->{name} =~ tr/A-Za-z0-9_//) and !defined ($_->{macro}) and !defined ($_->{value}) and !defined ($_->{default}) and !defined ($_->{pre}) and !defined ($_->{post}) and !defined ($_->{def_pre}) and !defined ($_->{def_post}) and !defined ($_->{weight})) { # It's the default type, and the name consists only of A-Za-z0-9_ push @simple, $_->{name}; } else { push @complex, $_; } } if (!defined $declare_types) { # Do they pass in any types we weren't already using? foreach (keys %$what) { next if $used_types{$_}; $declare_types++; # Found one in $what that wasn't used. last; # And one is enough to terminate this loop } } if ($declare_types) { $result = $indent . 'my $types = {map {($_, 1)} qw(' . join (" ", sort keys %$what) . ")};\n"; } local $Text::Wrap::huge = 'overflow'; local $Text::Wrap::columns = 80; $result .= wrap ($indent . "my \@names = (qw(", $indent . " ", join (" ", sort @simple) . ")"); if (@complex) { foreach my $item (sort {$a->{name} cmp $b->{name}} @complex) { my $name = perl_stringify $item->{name}; my $line = ",\n$indent {name=>\"$name\""; $line .= ", type=>\"$item->{type}\"" if defined $item->{type}; foreach my $thing (qw (macro value default pre post def_pre def_post)) { my $value = $item->{$thing}; if (defined $value) { if (ref $value) { $line .= ", $thing=>[\"" . join ('", "', map {perl_stringify $_} @$value) . '"]'; } else { $line .= ", $thing=>\"" . perl_stringify($value) . "\""; } } } $line .= "}"; # Ensure that the enclosing C comment doesn't end # by turning */ into *" . "/ $line =~ s!\*\/!\*" . "/!gs; # gcc -Wall doesn't like finding /* inside a comment $line =~ s!\/\*!/" . "\*!gs; $result .= $line; } } $result .= ");\n"; $result; } =item assign arg_hashref, VALUE... A method to return a suitable assignment clause. If I is aggregate (eg I expects both pointer and length) then there should be multiple Is for the components. I
 and I if defined give snippets
of C code to proceed and follow the assignment. I
 will be at the start
of a block, so variables may be defined in it.

=cut
# Hmm. value undef to to NOTDEF? value () to do NOTFOUND?

sub assign {
  my $self = shift;
  my $args = shift;
  my ($indent, $type, $pre, $post, $item)
      = @{$args}{qw(indent type pre post item)};
  $post ||= '';
  my $clause;
  my $close;
  if ($pre) {
    chomp $pre;
    $close = "$indent}\n";
    $clause = $indent . "{\n";
    $indent .= "  ";
    $clause .= "$indent$pre";
    $clause .= ";" unless $pre =~ /;$/;
    $clause .= "\n";
  }
  confess "undef \$type" unless defined $type;
  confess "Can't generate code for type $type"
    unless $self->valid_type($type);

  $clause .= join '', map {"$indent$_\n"}
    $self->assignment_clause_for_type({type=>$type,item=>$item}, @_);
  chomp $post;
  if (length $post) {
    $clause .= "$post";
    $clause .= ";" unless $post =~ /;$/;
    $clause .= "\n";
  }
  my $return = $self->return_statement_for_type($type);
  $clause .= "$indent$return\n" if defined $return;
  $clause .= $close if $close;
  return $clause;
}

=item return_clause arg_hashref, ITEM

A method to return a suitable C<#ifdef> clause. I is a hashref
(as passed to C and C. I is the number
of spaces to indent, defaulting to 6.

=cut

sub return_clause {

##ifdef thingy
#      *iv_return = thingy;
#      return PERL_constant_ISIV;
##else
#      return PERL_constant_NOTDEF;
##endif
  my ($self, $args, $item) = @_;
  my $indent = $args->{indent};

  my ($name, $value, $default, $pre, $post, $def_pre, $def_post, $type)
    = @$item{qw (name value default pre post def_pre def_post type)};
  $value = $name unless defined $value;
  my $macro = $self->macro_from_item($item);
  $indent = ' ' x ($indent || 6);
  unless (defined $type) {
    # use Data::Dumper; print STDERR Dumper ($item);
    confess "undef \$type";
  }

  ##ifdef thingy
  my $clause = $self->macro_to_ifdef($macro);

  #      *iv_return = thingy;
  #      return PERL_constant_ISIV;
  $clause
    .= $self->assign ({indent=>$indent, type=>$type, pre=>$pre, post=>$post,
		       item=>$item}, ref $value ? @$value : $value);

  if (defined $macro && $macro ne "" && $macro ne "1") {
    ##else
    $clause .= "#else\n";

    #      return PERL_constant_NOTDEF;
    if (!defined $default) {
      my $notdef = $self->return_statement_for_notdef();
      $clause .= "$indent$notdef\n" if defined $notdef;
    } else {
      my @default = ref $default ? @$default : $default;
      $type = shift @default;
      $clause .= $self->assign ({indent=>$indent, type=>$type, pre=>$pre,
				 post=>$post, item=>$item}, @default);
    }
  }
  ##endif
  $clause .= $self->macro_to_endif($macro);

  return $clause;
}

sub match_clause {
  # $offset defined if we have checked an offset.
  my ($self, $args, $item) = @_;
  my ($offset, $indent) = @{$args}{qw(checked_at indent)};
  $indent = ' ' x ($indent || 4);
  my $body = '';
  my ($no, $yes, $either, $name, $inner_indent);
  if (ref $item eq 'ARRAY') {
    ($yes, $no) = @$item;
    $either = $yes || $no;
    confess "$item is $either expecting hashref in [0] || [1]"
      unless ref $either eq 'HASH';
    $name = $either->{name};
  } else {
    confess "$item->{name} has utf8 flag '$item->{utf8}', should be false"
      if $item->{utf8};
    $name = $item->{name};
    $inner_indent = $indent;
  }

  $body .= $self->memEQ_clause ({name => $name, checked_at => $offset,
				 indent => length $indent});
  # If we've been presented with an arrayref for $item, then the user string
  # contains in the range 128-255, and we need to check whether it was utf8
  # (or not).
  # In the worst case we have two named constants, where one's name happens
  # encoded in UTF8 happens to be the same byte sequence as the second's
  # encoded in (say) ISO-8859-1.
  # In this case, $yes and $no both have item hashrefs.
  if ($yes) {
    $body .= $indent . "  if (" . $self->is_utf8_param . ") {\n";
  } elsif ($no) {
    $body .= $indent . "  if (!" . $self->is_utf8_param . ") {\n";
  }
  if ($either) {
    $body .= $self->return_clause ({indent=>4 + length $indent}, $either);
    if ($yes and $no) {
      $body .= $indent . "  } else {\n";
      $body .= $self->return_clause ({indent=>4 + length $indent}, $no);
    }
    $body .= $indent . "  }\n";
  } else {
    $body .= $self->return_clause ({indent=>2 + length $indent}, $item);
  }
  $body .= $indent . "}\n";
}


=item switch_clause arg_hashref, NAMELEN, ITEMHASH, ITEM...

An internal method to generate a suitable C clause, called by
C Is are in the hash ref format as given in the description
of C, and must all have the names of the same length, given by
I.  I is a reference to a hash, keyed by name, values being
the hashrefs in the I list.  (No parameters are modified, and there can
be keys in the I that are not in the list of Is without
causing problems - the hash is passed in to save generating it afresh for
each call).

=cut

sub switch_clause {
  my ($self, $args, $namelen, $items, @items) = @_;
  my ($indent, $comment) = @{$args}{qw(indent comment)};
  $indent = ' ' x ($indent || 2);

  local $Text::Wrap::huge = 'overflow';
  local $Text::Wrap::columns = 80;

  my @names = sort map {$_->{name}} @items;
  my $leader = $indent . '/* ';
  my $follower = ' ' x length $leader;
  my $body = $indent . "/* Names all of length $namelen.  */\n";
  if (defined $comment) {
    $body = wrap ($leader, $follower, $comment) . "\n";
    $leader = $follower;
  }
  my @safe_names = @names;
  foreach (@safe_names) {
    confess sprintf "Name '$_' is length %d, not $namelen", length
      unless length == $namelen;
    # Argh. 5.6.1
    # next unless tr/A-Za-z0-9_//c;
    next if tr/A-Za-z0-9_// == length;
    $_ = '"' . perl_stringify ($_) . '"';
    # Ensure that the enclosing C comment doesn't end
    # by turning */  into *" . "/
    s!\*\/!\*"."/!gs;
    # gcc -Wall doesn't like finding /* inside a comment
    s!\/\*!/"."\*!gs;
  }
  $body .= wrap ($leader, $follower, join (" ", @safe_names) . " */") . "\n";
  # Figure out what to switch on.
  # (RMS, Spread of jump table, Position, Hashref)
  my @best = (1e38, ~0);
  # Prefer the last character over the others. (As it lets us shorten the
  # memEQ clause at no cost).
  foreach my $i ($namelen - 1, 0 .. ($namelen - 2)) {
    my ($min, $max) = (~0, 0);
    my %spread;
    if (is_perl56) {
      # Need proper Unicode preserving hash keys for bytes in range 128-255
      # here too, for some reason. grr 5.6.1 yet again.
      tie %spread, 'ExtUtils::Constant::Aaargh56Hash';
    }
    foreach (@names) {
      my $char = substr $_, $i, 1;
      my $ord = ord $char;
      confess "char $ord is out of range" if $ord > 255;
      $max = $ord if $ord > $max;
      $min = $ord if $ord < $min;
      push @{$spread{$char}}, $_;
      # warn "$_ $char";
    }
    # I'm going to pick the character to split on that minimises the root
    # mean square of the number of names in each case. Normally this should
    # be the one with the most keys, but it may pick a 7 where the 8 has
    # one long linear search. I'm not sure if RMS or just sum of squares is
    # actually better.
    # $max and $min are for the tie-breaker if the root mean squares match.
    # Assuming that the compiler may be building a jump table for the
    # switch() then try to minimise the size of that jump table.
    # Finally use < not <= so that if it still ties the earliest part of
    # the string wins. Because if that passes but the memEQ fails, it may
    # only need the start of the string to bin the choice.
    # I think. But I'm micro-optimising. :-)
    # OK. Trump that. Now favour the last character of the string, before the
    # rest.
    my $ss;
    $ss += @$_ * @$_ foreach values %spread;
    my $rms = sqrt ($ss / keys %spread);
    if ($rms < $best[0] || ($rms == $best[0] && ($max - $min) < $best[1])) {
      @best = ($rms, $max - $min, $i, \%spread);
    }
  }
  confess "Internal error. Failed to pick a switch point for @names"
    unless defined $best[2];
  # use Data::Dumper; print Dumper (@best);
  my ($offset, $best) = @best[2,3];
  $body .= $indent . "/* Offset $offset gives the best switch position.  */\n";

  my $do_front_chop = $offset == 0 && $namelen > 2;
  if ($do_front_chop) {
    $body .= $indent . "switch (*" . $self->name_param() . "++) {\n";
  } else {
    $body .= $indent . "switch (" . $self->name_param() . "[$offset]) {\n";
  }
  foreach my $char (sort keys %$best) {
    confess sprintf "'$char' is %d bytes long, not 1", length $char
      if length ($char) != 1;
    confess sprintf "char %#X is out of range", ord $char if ord ($char) > 255;
    $body .= $indent . "case '" . C_stringify ($char) . "':\n";
    foreach my $thisone (sort {
	# Deal with the case of an item actually being an array ref to 1 or 2
	# hashrefs. Don't assign to $a or $b, as they're aliases to the orignal
	my $l = ref $a eq 'ARRAY' ? ($a->[0] || $->[1]) : $a;
	my $r = ref $b eq 'ARRAY' ? ($b->[0] || $->[1]) : $b;
	# Sort by weight first
	($r->{weight} || 0) <=> ($l->{weight} || 0)
	    # Sort equal weights by name
	    or $l->{name} cmp $r->{name}}
			 # If this looks evil, maybe it is.  $items is a
			 # hashref, and we're doing a hash slice on it
			 @{$items}{@{$best->{$char}}}) {
      # warn "You are here";
      if ($do_front_chop) {
        $body .= $self->match_clause ({indent => 2 + length $indent,
				       checked_at => \$char}, $thisone);
      } else {
        $body .= $self->match_clause ({indent => 2 + length $indent,
				       checked_at => $offset}, $thisone);
      }
    }
    $body .= $indent . "  break;\n";
  }
  $body .= $indent . "}\n";
  return $body;
}

sub C_constant_return_type {
  "static int";
}

sub C_constant_prefix_param {
  '';
}

sub C_constant_prefix_param_defintion {
  '';
}

sub name_param_definition {
  "const char *" . $_[0]->name_param;
}

sub namelen_param {
  'len';
}

sub namelen_param_definition {
  'size_t ' . $_[0]->namelen_param;
}

sub C_constant_other_params {
  '';
}

sub C_constant_other_params_defintion {
  '';
}

=item params WHAT

An "internal" method, subject to change, currently called to allow an
overriding class to cache information that will then be passed into all
the C<*param*> calls. (Yes, having to read the source to make sense of this is
considered a known bug). I is be a hashref of types the constant
function will return. In ExtUtils::Constant::XS this method is used to
returns a hashref keyed IV NV PV SV to show which combination of pointers will
be needed in the C argument list generated by
C_constant_other_params_definition and C_constant_other_params

=cut

sub params {
  '';
}


=item dogfood arg_hashref, ITEM...

An internal function to generate the embedded perl code that will regenerate
the constant subroutines.  Parameters are the same as for C_constant.

Currently the base class does nothing and returns an empty string.

=cut

sub dogfood {
  ''
}

=item normalise_items args, default_type, seen_types, seen_items, ITEM...

Convert the items to a normalised form. For 8 bit and Unicode values converts
the item to an array of 1 or 2 items, both 8 bit and UTF-8 encoded.

=cut

sub normalise_items
{
    my $self = shift;
    my $args = shift;
    my $default_type = shift;
    my $what = shift;
    my $items = shift;
    my @new_items;
    foreach my $orig (@_) {
	my ($name, $item);
      if (ref $orig) {
        # Make a copy which is a normalised version of the ref passed in.
        $name = $orig->{name};
        my ($type, $macro, $value) = @$orig{qw (type macro value)};
        $type ||= $default_type;
        $what->{$type} = 1;
        $item = {name=>$name, type=>$type};

        undef $macro if defined $macro and $macro eq $name;
        $item->{macro} = $macro if defined $macro;
        undef $value if defined $value and $value eq $name;
        $item->{value} = $value if defined $value;
        foreach my $key (qw(default pre post def_pre def_post weight
			    not_constant)) {
          my $value = $orig->{$key};
          $item->{$key} = $value if defined $value;
          # warn "$key $value";
        }
      } else {
        $name = $orig;
        $item = {name=>$name, type=>$default_type};
        $what->{$default_type} = 1;
      }
      warn +(ref ($self) || $self)
	. "doesn't know how to handle values of type $_ used in macro $name"
	  unless $self->valid_type ($item->{type});
      # tr///c is broken on 5.6.1 for utf8, so my original tr/\0-\177//c
      # doesn't work. Upgrade to 5.8
      # if ($name !~ tr/\0-\177//c || $] < 5.005_50) {
      if ($name =~ tr/\0-\177// == length $name || $] < 5.005_50
	 || $args->{disable_utf8_duplication}) {
        # No characters outside 7 bit ASCII.
        if (exists $items->{$name}) {
          die "Multiple definitions for macro $name";
        }
        $items->{$name} = $item;
      } else {
        # No characters outside 8 bit. This is hardest.
        if (exists $items->{$name} and ref $items->{$name} ne 'ARRAY') {
          confess "Unexpected ASCII definition for macro $name";
        }
        # Again, 5.6.1 tr broken, so s/5\.6.*/5\.8\.0/;
        # if ($name !~ tr/\0-\377//c) {
        if ($name =~ tr/\0-\377// == length $name) {
#          if ($] < 5.007) {
#            $name = pack "C*", unpack "U*", $name;
#          }
          $item->{utf8} = 'no';
          $items->{$name}[1] = $item;
          push @new_items, $item;
          # Copy item, to create the utf8 variant.
          $item = {%$item};
        }
        # Encode the name as utf8 bytes.
        unless (is_perl56) {
          utf8::encode($name);
        } else {
#          warn "Was >$name< " . length ${name};
          $name = pack 'C*', unpack 'C*', $name . pack 'U*';
#          warn "Now '${name}' " . length ${name};
        }
        if ($items->{$name}[0]) {
          die "Multiple definitions for macro $name";
        }
        $item->{utf8} = 'yes';
        $item->{name} = $name;
        $items->{$name}[0] = $item;
        # We have need for the utf8 flag.
        $what->{''} = 1;
      }
      push @new_items, $item;
    }
    @new_items;
}

=item C_constant arg_hashref, ITEM...

A function that returns a B of C subroutine definitions that return
the value and type of constants when passed the name by the XS wrapper.
I gives a list of constant names. Each can either be a string,
which is taken as a C macro name, or a reference to a hash with the following
keys

=over 8

=item name

The name of the constant, as seen by the perl code.

=item type

The type of the constant (I, I etc)

=item value

A C expression for the value of the constant, or a list of C expressions if
the type is aggregate. This defaults to the I if not given.

=item macro

The C pre-processor macro to use in the C<#ifdef>. This defaults to the
I, and is mainly used if I is an C. If a reference an
array is passed then the first element is used in place of the C<#ifdef>
line, and the second element in place of the C<#endif>. This allows
pre-processor constructions such as

    #if defined (foo)
    #if !defined (bar)
    ...
    #endif
    #endif

to be used to determine if a constant is to be defined.

A "macro" 1 signals that the constant is always defined, so the C<#if>/C<#endif>
test is omitted.

=item default

Default value to use (instead of Cing with "your vendor has not
defined...") to return if the macro isn't defined. Specify a reference to
an array with type followed by value(s).

=item pre

C code to use before the assignment of the value of the constant. This allows
you to use temporary variables to extract a value from part of a C
and return this as I. This C code is places at the start of a block,
so you can declare variables in it.

=item post

C code to place between the assignment of value (to a temporary) and the
return from the function. This allows you to clear up anything in I
.
Rarely needed.

=item def_pre

=item def_post

Equivalents of I
 and I for the default value.

=item utf8

Generated internally. Is zero or undefined if name is 7 bit ASCII,
"no" if the name is 8 bit (and so should only match if SvUTF8() is false),
"yes" if the name is utf8 encoded.

The internals automatically clone any name with characters 128-255 but none
256+ (ie one that could be either in bytes or utf8) into a second entry
which is utf8 encoded.

=item weight

Optional sorting weight for names, to determine the order of
linear testing when multiple names fall in the same case of a switch clause.
Higher comes earlier, undefined defaults to zero.

=back

In the argument hashref, I is the name of the package, and is only
used in comments inside the generated C code. I defaults to
C if undefined.

I is the type returned by Cs that don't specify their
type. It defaults to the value of C. I should be given
either as a comma separated list of types that the C subroutine I
will generate or as a reference to a hash. I will be added to
the list if not present, as will any types given in the list of Is. The
resultant list should be the same list of types that C is
given. [Otherwise C and C may differ in the number of
parameters to the constant function. I is currently unused and
ignored. In future it may be used to pass in information used to change the C
indentation style used.]  The best way to maintain consistency is to pass in a
hash reference and let this function update it.

I governs when child functions of I are generated.  If there
are I or more Is with the same length of name, then the code
to switch between them is placed into a function named I_I, for
example C for names 5 characters long.  The default I is
3.  A single C is always inlined.

=cut

# The parameter now BREAKOUT was previously documented as:
#
# I if defined signals that all the Is of the Is are of
# this length, and that the constant name passed in by perl is checked and
# also of this length. It is used during recursion, and should be C
# unless the caller has checked all the lengths during code generation, and
# the generated subroutine is only to be called with a name of this length.
#
# As you can see it now performs this function during recursion by being a
# scalar reference.

sub C_constant {
  my ($self, $args, @items) = @_;
  my ($package, $subname, $default_type, $what, $indent, $breakout) =
    @{$args}{qw(package subname default_type types indent breakout)};
  $package ||= 'Foo';
  $subname ||= 'constant';
  # I'm not using this. But a hashref could be used for full formatting without
  # breaking this API
  # $indent ||= 0;

  my ($namelen, $items);
  if (ref $breakout) {
    # We are called recursively. We trust @items to be normalised, $what to
    # be a hashref, and pinch %$items from our parent to save recalculation.
    ($namelen, $items) = @$breakout;
  } else {
    $items = {};
    if (is_perl56) {
      # Need proper Unicode preserving hash keys.
      require ExtUtils::Constant::Aaargh56Hash;
      tie %$items, 'ExtUtils::Constant::Aaargh56Hash';
    }
    $breakout ||= 3;
    $default_type ||= $self->default_type();
    if (!ref $what) {
      # Convert line of the form IV,UV,NV to hash
      $what = {map {$_ => 1} split /,\s*/, ($what || '')};
      # Figure out what types we're dealing with, and assign all unknowns to the
      # default type
    }
    @items = $self->normalise_items ({}, $default_type, $what, $items, @items);
    # use Data::Dumper; print Dumper @items;
  }
  my $params = $self->params ($what);

  # Probably "static int"
  my ($body, @subs);
  $body = $self->C_constant_return_type($params) . "\n$subname ("
    # Eg "pTHX_ "
    . $self->C_constant_prefix_param_defintion($params)
      # Probably "const char *name"
      . $self->name_param_definition($params);
  # Something like ", STRLEN len"
  $body .= ", " . $self->namelen_param_definition($params)
    unless defined $namelen;
  $body .= $self->C_constant_other_params_defintion($params);
  $body .= ") {\n";

  if (defined $namelen) {
    # We are a child subroutine. Print the simple description
    my $comment = 'When generated this function returned values for the list'
      . ' of names given here.  However, subsequent manual editing may have'
        . ' added or removed some.';
    $body .= $self->switch_clause ({indent=>2, comment=>$comment},
				   $namelen, $items, @items);
  } else {
    # We are the top level.
    $body .= "  /* Initially switch on the length of the name.  */\n";
    $body .= $self->dogfood ({package => $package, subname => $subname,
			      default_type => $default_type, what => $what,
			      indent => $indent, breakout => $breakout},
			     @items);
    $body .= '  switch ('.$self->namelen_param().") {\n";
    # Need to group names of the same length
    my @by_length;
    foreach (@items) {
      push @{$by_length[length $_->{name}]}, $_;
    }
    foreach my $i (0 .. $#by_length) {
      next unless $by_length[$i];	# None of this length
      $body .= "  case $i:\n";
      if (@{$by_length[$i]} == 1) {
        my $only_thing = $by_length[$i]->[0];
        if ($only_thing->{utf8}) {
          if ($only_thing->{utf8} eq 'yes') {
            # With utf8 on flag item is passed in element 0
            $body .= $self->match_clause (undef, [$only_thing]);
          } else {
            # With utf8 off flag item is passed in element 1
            $body .= $self->match_clause (undef, [undef, $only_thing]);
          }
        } else {
          $body .= $self->match_clause (undef, $only_thing);
        }
      } elsif (@{$by_length[$i]} < $breakout) {
        $body .= $self->switch_clause ({indent=>4},
				       $i, $items, @{$by_length[$i]});
      } else {
        # Only use the minimal set of parameters actually needed by the types
        # of the names of this length.
        my $what = {};
        foreach (@{$by_length[$i]}) {
          $what->{$_->{type}} = 1;
          $what->{''} = 1 if $_->{utf8};
        }
        $params = $self->params ($what);
        push @subs, $self->C_constant ({package=>$package,
					subname=>"${subname}_$i",
					default_type => $default_type,
					types => $what, indent => $indent,
					breakout => [$i, $items]},
				       @{$by_length[$i]});
        $body .= "    return ${subname}_$i ("
	  # Eg "aTHX_ "
	  . $self->C_constant_prefix_param($params)
	    # Probably "name"
	    . $self->name_param($params);
	$body .= $self->C_constant_other_params($params);
        $body .= ");\n";
      }
      $body .= "    break;\n";
    }
    $body .= "  }\n";
  }
  my $notfound = $self->return_statement_for_notfound();
  $body .= "  $notfound\n" if $notfound;
  $body .= "}\n";
  return (@subs, $body);
}

1;
__END__

=back

=head1 BUGS

Not everything is documented yet.

Probably others.

=head1 AUTHOR

Nicholas Clark  based on the code in C by Larry Wall and
others
usr/share/perl5/vendor_perl/TAP/Base.pm000064400000004375152535571450013706 0ustar00package TAP::Base;

use strict;
use warnings;

use base 'TAP::Object';

=head1 NAME

TAP::Base - Base class that provides common functionality to L
and L

=head1 VERSION

Version 3.42

=cut

our $VERSION = '3.42';

use constant GOT_TIME_HIRES => do {
    eval 'use Time::HiRes qw(time);';
    $@ ? 0 : 1;
};

=head1 SYNOPSIS

    package TAP::Whatever;

    use base 'TAP::Base';

    # ... later ...
    
    my $thing = TAP::Whatever->new();
    
    $thing->callback( event => sub {
        # do something interesting
    } );

=head1 DESCRIPTION

C provides callback management.

=head1 METHODS

=head2 Class Methods

=cut

sub _initialize {
    my ( $self, $arg_for, $ok_callback ) = @_;

    my %ok_map = map { $_ => 1 } @$ok_callback;

    $self->{ok_callbacks} = \%ok_map;

    if ( my $cb = delete $arg_for->{callbacks} ) {
        while ( my ( $event, $callback ) = each %$cb ) {
            $self->callback( $event, $callback );
        }
    }

    return $self;
}

=head3 C

Install a callback for a named event.

=cut

sub callback {
    my ( $self, $event, $callback ) = @_;

    my %ok_map = %{ $self->{ok_callbacks} };

    $self->_croak('No callbacks may be installed')
      unless %ok_map;

    $self->_croak( "Callback $event is not supported. Valid callbacks are "
          . join( ', ', sort keys %ok_map ) )
      unless exists $ok_map{$event};

    push @{ $self->{code_for}{$event} }, $callback;

    return;
}

sub _has_callbacks {
    my $self = shift;
    return keys %{ $self->{code_for} } != 0;
}

sub _callback_for {
    my ( $self, $event ) = @_;
    return $self->{code_for}{$event};
}

sub _make_callback {
    my $self  = shift;
    my $event = shift;

    my $cb = $self->_callback_for($event);
    return unless defined $cb;
    return map { $_->(@_) } @$cb;
}

=head3 C

Return the current time using Time::HiRes if available.

=cut

sub get_time { return time() }

=head3 C

Return true if the time returned by get_time is high resolution (i.e. if Time::HiRes is available).

=cut

sub time_is_hires { return GOT_TIME_HIRES }

=head3 C

Return array reference of the four-element list of CPU seconds,
as with L.

=cut

sub get_times { return [ times() ] }

1;
usr/lib64/perl5/vendor_perl/DBI/Gofer/Transport/Base.pm000064400000011664152536603640016531 0ustar00package DBI::Gofer::Transport::Base;

#   $Id: Base.pm 12536 2009-02-24 22:37:09Z Tim $
#
#   Copyright (c) 2007, Tim Bunce, Ireland
#
#   You may distribute under the terms of either the GNU General Public
#   License or the Artistic License, as specified in the Perl README file.

use strict;
use warnings;

use DBI;

use base qw(DBI::Util::_accessor);

use DBI::Gofer::Serializer::Storable;
use DBI::Gofer::Serializer::DataDumper;

our $VERSION = "0.012537";

__PACKAGE__->mk_accessors(qw(
    trace
    keep_meta_frozen
    serializer_obj
));


# see also $ENV{DBI_GOFER_TRACE} in DBI::Gofer::Execute
sub _init_trace { (split(/=/,$ENV{DBI_GOFER_TRACE}||0))[0] }


sub new {
    my ($class, $args) = @_;
    $args->{trace} ||= $class->_init_trace;
    $args->{serializer_obj} ||= DBI::Gofer::Serializer::Storable->new();
    my $self = bless {}, $class;
    $self->$_( $args->{$_} ) for keys %$args;
    $self->trace_msg("$class->new({ @{[ %$args ]} })\n") if $self->trace;
    return $self;
}

my $packet_header_text  = "GoFER1:";
my $packet_header_regex = qr/^GoFER(\d+):/;


sub _freeze_data {
    my ($self, $data, $serializer, $skip_trace) = @_;
    my $frozen = eval {
        $self->_dump("freezing $self->{trace} ".ref($data), $data)
            if !$skip_trace and $self->trace;

        local $data->{meta}; # don't include meta in serialization
	$serializer ||= $self->{serializer_obj};
        my ($data, $deserializer_class)  = $serializer->serialize($data);

        $packet_header_text . $data;
    };
    if ($@) {
        chomp $@;
        die "Error freezing ".ref($data)." object: $@";
    }

    # stash the frozen data into the data structure itself
    # to make life easy for the client caching code in DBD::Gofer::Transport::Base
    $data->{meta}{frozen} = $frozen if $self->keep_meta_frozen;

    return $frozen;
}
# public aliases used by subclasses
*freeze_request  = \&_freeze_data;
*freeze_response = \&_freeze_data;


sub _thaw_data {
    my ($self, $frozen_data, $serializer, $skip_trace) = @_;
    my $data;
    eval {
        # check for and extract our gofer header and the info it contains
        (my $frozen = $frozen_data) =~ s/$packet_header_regex//o
            or die "does not have gofer header\n";
        my ($t_version) = $1;
	$serializer ||= $self->{serializer_obj};
        $data = $serializer->deserialize($frozen);
        die ref($serializer)."->deserialize didn't return a reference"
            unless ref $data;
        $data->{_transport}{version} = $t_version;

        $data->{meta}{frozen} = $frozen_data if $self->keep_meta_frozen;
    };
    if ($@) {
        chomp(my $err = $@);
        # remove extra noise from Storable
        $err =~ s{ at \S+?/Storable.pm \(autosplit into \S+?/Storable/thaw.al\) line \d+(, \S+ line \d+)?}{};
        my $msg = sprintf "Error thawing: %s (data=%s)", $err, DBI::neat($frozen_data,50);
        Carp::cluck("$msg, pid $$ stack trace follows:"); # XXX if $self->trace;
        die $msg;
    }
    $self->_dump("thawing $self->{trace} ".ref($data), $data)
        if !$skip_trace and $self->trace;

    return $data;
}
# public aliases used by subclasses
*thaw_request  = \&_thaw_data;
*thaw_response = \&_thaw_data;


# this should probably live in the request and response classes
# and the tace level passed in
sub _dump {
    my ($self, $label, $data) = @_;

    # don't dump the binary
    local $data->{meta}{frozen} if $data->{meta} && $data->{meta}{frozen};

    my $trace_level = $self->trace;
    my $summary;
    if ($trace_level >= 4) {
        require Data::Dumper;
        local $Data::Dumper::Indent    = 1;
        local $Data::Dumper::Terse     = 1;
        local $Data::Dumper::Useqq     = 0;
        local $Data::Dumper::Sortkeys  = 1;
        local $Data::Dumper::Quotekeys = 0;
        local $Data::Dumper::Deparse   = 0;
        local $Data::Dumper::Purity    = 0;
        $summary = Data::Dumper::Dumper($data);
    }
    elsif ($trace_level >= 2) {
        $summary = eval { $data->summary_as_text } || $@ || "no summary available\n";
    }
    else {
        $summary = eval { $data->outline_as_text."\n" } || $@ || "no summary available\n";
    }
    $self->trace_msg("$label: $summary");
}


sub trace_msg {
    my ($self, $msg, $min_level) = @_;
    $min_level = 1 unless defined $min_level;
    # transport trace level can override DBI's trace level
    $min_level = 0 if $self->trace >= $min_level;
    return DBI->trace_msg("gofer ".$msg, $min_level);
}

1;

=head1 NAME

DBI::Gofer::Transport::Base - Base class for Gofer transports

=head1 DESCRIPTION

This is the base class for server-side Gofer transports.

It's also the base class for the client-side base class L.

This is an internal class.

=head1 AUTHOR

Tim Bunce, L

=head1 LICENCE AND COPYRIGHT

Copyright (c) 2007, Tim Bunce, Ireland. All rights reserved.

This module is free software; you can redistribute it and/or
modify it under the same terms as Perl itself. See L.

=cut