/ 
Devel-Size-Report-0.13
River stage zero No dependents
/ Devel::Size::Report

NAME

Devel::Size::Report - generate a size report for all elements in a structure

SYNOPSIS

use Devel::Size::Report qw/report_size/;

my $a = [ \8, \*STDIN, 7,
          [ 1, 2, 3,
            { a => 'b',
              size => 12.2,
              h => ['a']
            },
          'rrr'
          ]
        ];
print report_size($a, { indent => "  " } );

This will print something like this:

Size report v0.08 for 'ARRAY(0x8145e6c)':
  Array 886 bytes (overhead: 100 bytes, 11.29%)
    Scalar Ref 32 bytes (overhead: 16 bytes, 50.00%)
      Read-Only Scalar 16 bytes
    Glob 266 bytes
    Scalar 16 bytes
    Array 472 bytes (overhead: 88 bytes, 18.64%)
      Scalar 16 bytes
      Scalar 16 bytes
      Scalar 16 bytes
      Hash 308 bytes (overhead: 180 bytes, 58.44%)
        'h' => Array 82 bytes (overhead: 56 bytes, 68.29%)
          Scalar 26 bytes
        'a' => Scalar 26 bytes
        'size' => Scalar 20 bytes
      Scalar 28 bytes
Total: 886 bytes in 15 elements

EXPORTS

Nothing per default, but can export the following per request:

	report_size
	track_size
	track_sizes
	hide_tracks
	element_type
	entries_per_element

	S_SCALAR
	S_HASH
	S_ARRAY
	S_GLOB
	S_UNKNOWN
	S_CODE
	S_LVALUE
	S_REGEXP
  	S_CYCLE
	S_DOUBLE

	SF_KEY
	SF_REF
	SF_REF
	SF_WEAK
	SF_RO
	SF_VSTRING
	SF_DUAL

DESCRIPTION

Devel::Size can only report the size of a single element or the total size of a structure (array, hash etc). This module enhances Devel::Size by giving you the ability to generate a full size report for each element in a structure.

You have full control over how the generated text report looks like, and where you want to output it. In addition, the method track_size allows you to get at the raw data that is used to generate the report for even more flexibility.

METHODS

report_size

my $record = report_size( $reference, $options ) . "\n";
print $record;

Walks the given reference recursively and returns text tree describing the size of each element. $options is a hash, containing the following optional keys:

	names	  ref to HASH mapping the types to names
		  This should map S_Scalar to something like "Scalar" etc
	indent	  string to indent different levels with, default is '  '
	left	  indent all text with this at the left side, default is ''
	inner	  indent inner text with this at the left side, default is '  '
	total	  if true, a total size will be printed as last line
	bytes	  name of the size unit, defaults to 'bytes'
	head	  header string, default 'Size report for'
		  Set to '' to supress header completely	
  	overhead  Format string for the overhead, first size in bytes, then
		  the bytes string (see above) and then the percentage.
		  The default is:
		  " (overhead: %i%s, %0.2f%%)"
	addr 	  if true, for each element the memory address will be output
	class	  if true, show the class each element was blessed in
	terse	  if true, details for elements will be supressed, e.g. you
		  will only get the header, total and summary (if requested)
	summary   if true, print a table summing up the memory details on a
		  per-class basis
	doubles	  If true, hash keys and array elemts that point to the same
		  thing in memory will be reported. Default is off, since it
		  saves memory. Note that you will usually only get double
		  entries in hashes and array by using Array::RefElem's methods
		  or similiar hacks/tricks.

entries_per_element

my $entries = entries_per_element();

Returns the number of entries per element that track_size() will generate.

track_sizes

$elements = track_sizes( $reference, $level);

Walk the given scalar or reference recursively and returns a ref to an array, containing entries_per_element entries for each element in the structure pointed to by $reference. $reference can also be a plain scalar.

The entries for each element are currently:

level	  the indent level
type	  the type of the element, S_SCALAR, S_HASH, S_ARRAY etc
	  if (type & SF_KEY) != 0, the element is a member of a hash
size	  size in bytes of the element
overhead  if the element is an ARRAY or HASH, contains the overhead
	  in bytes (size - sum(size of all elements)).
name	  if type & SF_KEY != 0, this contains the name of the hash key
addr	  memory address of the element
class	  classname that the element was blessed into, or ''

track_size

@elements = track_size( $reference, $level);

Works just like track_sizes, but for backward compatibility reasons returns an array with the results.

hide_tracks

hide_tracks();

Releases the memory consumed by a call to track_size or track_sizes.

type

$type_number = type($type_name);

Maps a type name (like 'SCALAR') to a type number (lilke S_SCALAR).

element_type

$type_name = element_type($type_nr);

Maps a type number (like S_SCALAR) to a type name (lilke 'SCALAR').

WRAP YOUR OWN

If you want to create your own report with different formattings, please use track_size and create a report out of the data you get back from it. Look at the source code for report_size how to do this - it is easy!

CAVEATS

  • The limitations of Devel::Size also apply to this module. This means that CODE refs and other "obscure" things might show wrong sizes, or unknown overhead. In addition, some sizes might be reported wrong.

  • A string representation of the passed argument will be inserted when generating a report with a header.

    If the passed argument is an object with overloaded magic, then the routine for stringification will be triggered. If this routine does actually modify the object (for instance, Math::String objects cache their string form upon first call to stringification, thus modifying themselves), the reported size will be different from a first report without a header.

BUGS

threads

The results are currently stored in a global package var, so this is probably not threadsafe.

AUTHOR

(c) 2004, 2005, 2006, 2008 by Tels http://bloodgate.com