我看到了How can I redefine Perl class methods?,所以我想通过一个示例更好地了解它的工作方式。 (相关:How do I reference methods? - PerlMonks)

如果我有一个对象$obj,它是某些具有ClassClass::method的实例,则$obj也具有$obj->method;我认为在这种情况下,简化的内存布局将类似于以下内容(在LaTeX Table Generator中创建的表;对于下面的表,src here):

...也就是说,在(说)地址0x1000处,我们有$obj->method,它只是(以某种方式)指向实际的Class::method定义,即(说)在0x3500处。

假设我在主文件中的某个地方定义了sub,即somefunc(忽略表中的前缀$),其定义最终位于地址0x2000。

如果我仅对$obj实例“monkey-patch” /替换该方法,我希望内存布局看起来像这样(src here):

现在是这样的-在下面的示例中(东西的名称比上表中的唯一),我实际上想用在主文件Demo::My::Functions::test_me中定义的方法替换整个类方法repl_test_me。我真的不知道对地址有什么期望,所以我尝试使用%pprintf说明符来显示我认为在方法修补之前和之后的地址。代码输出如下:

$ perl cltest.pl
Starting Demo::Main...
Orig test_me!
1: DMFo: 8e63dc0 DMFo->test_me 8e916f8 DMF::test_me 8e916e8 repl_test_me 8e91748
Orig test_me!
-- CODE --
Subroutine Demo::My::Functions::test_me redefined at cltest.pl line 59.
Repl test_me!
Repl test_me!
2: DMFo: 8e63dc0 DMFo->test_me 8e916f8 DMF::test_me 8dfb708 repl_test_me 8dfb6c8

奇怪的是,即使是在主文件repl_test_me中定义的函数也更改了地址-不应该吗?

因此,我显然没有打印我认为是函数地址的内容-而且我认为,每个调用有两个打印输出,而我应该只有一个打印输出,这一事实也得到了证实。

如何更改下面的代码,以便打印出可以帮助我确认补丁/重载按预期发生的地址?

这是代码cltest.pl:
use v5.10.1;

package Demo::My::Functions;
$INC{'Demo/My/Functions.pm'} = 1;

use warnings;
use strict;

use base 'Class::Accessor';

__PACKAGE__->mk_accessors(qw(test_me));

sub test_me {
  my $self = shift;
  print("Orig test_me!\n");
  return 1;
}

sub test_also_me {
  my $self = shift;
  print("Orig test_also_me!\n");
  return 1;
}

sub new {
  my $class = shift;
  my $self = {};
  bless $self, $class;
  return $self;
}

############################################################
package Demo::Main;


use warnings;
use strict;
print("Starting Demo::Main...\n");

my $DMFA = Demo::My::Functions->new();

sub repl_test_me {
  my $self = shift;
  print("Repl test_me!\n");
  return 1;
}

# note: \&{$DMFA->test_me} calls!
printf("1: DMFo: %p DMFo->test_me %p DMF::test_me %p repl_test_me %p\n",
      $DMFA, \&{$DMFA->test_me}, \&{'Demo::My::Functions::test_me'}, \&repl_test_me
);

print("-- " . ref(\&{$DMFA->test_me}) . " --\n");

{
no strict 'refs';
#~ *Demo::My::Functions::test_me = sub {my $self = shift; print("ReplIN test_me!\n"); return 1; }; # OK
#~ *Demo::My::Functions::test_me = *repl_test_me; # overloads
*Demo::My::Functions::test_me = \&repl_test_me; # overloads
};

# test it:
$DMFA->test_me();

# output addr again:
printf("2: DMFo: %p DMFo->test_me %p DMF::test_me %p repl_test_me %p\n",
      $DMFA, \&{$DMFA->test_me}, \&{'Demo::My::Functions::test_me'}, \&repl_test_me
);

最佳答案

对象没有方法。类。一个对象只是一个变量,带有与之关联的包。

$ perl -MDevel::Peek -E'$o={}; Dump($o); bless($o); Dump($o); say \%main::'
SV = IV(0x26c2360) at 0x26c2370
  REFCNT = 1
  FLAGS = (ROK)
  RV = 0x269a978
  SV = PVHV(0x26a0400) at 0x269a978
    REFCNT = 1
    FLAGS = (SHAREKEYS)
    ARRAY = 0x0
    KEYS = 0
    FILL = 0
    MAX = 7
SV = IV(0x26c2360) at 0x26c2370
  REFCNT = 1
  FLAGS = (ROK)
  RV = 0x269a978
  SV = PVHV(0x26a0400) at 0x269a978
    REFCNT = 1
    FLAGS = (OBJECT,SHAREKEYS)
    STASH = 0x269a7f8   "main"
    ARRAY = 0x0
    KEYS = 0
    FILL = 0
    MAX = 7
HASH(0x269a7f8)    # Address of the main package.

所以真正的布局是
                +---------+
                |         |
                |         v
+-----------+   |   +-----------+
| Reference |   |   | 0x269a7f8 |
| 0x26c2370 |   |   | Package   |
+-----------+   |   +-----------+
      |         |         |
  References    |     Contains
      |         |         |
      v         |         v
+-----------+   |   +-----------+
| Hash      |   |   | 0xXXXXXXX |
| 0x269a978 |   |   | Method    |
+-----------+   |   +-----------+
      |         |
Blessed into    |
      |         |
      +---------+

更改程序包会影响使用该程序包的所有对象。
$ perl -E'
   sub f { "a" }
   my $o = bless({});
   say join " ", \&f, $o->can("f"), $o->f;
   *f = sub { "b" };
   say join " ", \&f, $o->can("f"), $o->f;
'
CODE(0x311c680) CODE(0x311c680) a
CODE(0x3126f60) CODE(0x3126f60) b

关于perl - 打印Perl对象方法的地址? (用于重新定义Perl类方法),我们在Stack Overflow上找到一个类似的问题:https://stackoverflow.com/questions/27206371/

10-11 22:23
查看更多