package Data::Dump;

use strict;
use vars qw(@EXPORT @EXPORT_OK $VERSION $DEBUG);
use subs qq(dump);

require Exporter;
*import = \&Exporter::import;
@EXPORT = qw(dd ddx);
@EXPORT_OK = qw(dump pp dumpf quote);

$VERSION = "1.17";
$DEBUG = 0;

use overload ();
use vars qw(%seen %refcnt @dump @fixup %require $TRY_BASE64 @FILTERS $INDENT);

$TRY_BASE64 = 50 unless defined $TRY_BASE64;
$INDENT = "  " unless defined $INDENT;

sub dump
{
    local %seen;
    local %refcnt;
    local %require;
    local @fixup;

    require Data::Dump::FilterContext if @FILTERS;

    my $name = "a";
    my @dump;

    for my $v (@_) {
	my $val = _dump($v, $name, [], tied($v));
	push(@dump, [$name, $val]);
    } continue {
	$name++;
    }

    my $out = "";
    if (%require) {
	for (sort keys %require) {
	    $out .= "require $_;\n";
	}
    }
    if (%refcnt) {
	# output all those with refcounts first
	for (@dump) {
	    my $name = $_->[0];
	    if ($refcnt{$name}) {
		$out .= "my \$$name = $_->[1];\n";
		undef $_->[1];
	    }
	}
	for (@fixup) {
	    $out .= "$_;\n";
	}
    }

    my $paren = (@dump != 1);
    $out .= "(" if $paren;
    $out .= format_list($paren, undef,
			map {defined($_->[1]) ? $_->[1] : "\$".$_->[0]}
			    @dump
		       );
    $out .= ")" if $paren;

    if (%refcnt || %require) {
	$out .= ";\n";
	$out =~ s/^/$INDENT/gm;
	$out = "do {\n$out}";
    }

    #use Data::Dumper;   print Dumper(\%refcnt);
    #use Data::Dumper;   print Dumper(\%seen);

    print STDERR "$out\n" unless defined wantarray;
    $out;
}

*pp = \&dump;

sub dd {
    print dump(@_), "\n";
}

sub ddx {
    my(undef, $file, $line) = caller;
    $file =~ s,.*[\\/],,;
    my $out = "$file:$line: " . dump(@_) . "\n";
    $out =~ s/^/# /gm;
    print $out;
}

sub dumpf {
    require Data::Dump::Filtered;
    goto &Data::Dump::Filtered::dump_filtered;
}

sub _dump
{
    my $ref  = ref $_[0];
    my $rval = $ref ? $_[0] : \$_[0];
    shift;

    my($name, $idx, $dont_remember, $pclass, $pidx) = @_;

    my($class, $type, $id);
    if (overload::StrVal($rval) =~ /^(?:([^=]+)=)?([A-Z]+)\(0x([^\)]+)\)$/) {
	$class = $1;
	$type  = $2;
	$id    = $3;
    } else {
	die "Can't parse " . overload::StrVal($rval);
    }
    if ($] < 5.008 && $type eq "SCALAR") {
	$type = "REF" if $ref eq "REF";
    }
    warn "\$$name(@$idx) $class $type $id ($ref)" if $DEBUG;

    my $out;
    my $comment;
    my $hide_keys;
    if (@FILTERS) {
	my $pself = "";
	$pself = fullname("self", [@$idx[$pidx..(@$idx - 1)]]) if $pclass;
	my $ctx = Data::Dump::FilterContext->new($rval, $class, $type, $ref, $pclass, $pidx, $idx);
	my @bless;
	for my $filter (@FILTERS) {
	    if (my $f = $filter->($ctx, $rval)) {
		if (my $v = $f->{object}) {
		    local @FILTERS;
		    $out = _dump($v, $name, $idx, 1);
		    $dont_remember++;
		}
		if (defined(my $c = $f->{bless})) {
		    push(@bless, $c);
		}
		if (my $c = $f->{comment}) {
		    $comment = $c;
		}
		if (defined(my $c = $f->{dump})) {
		    $out = $c;
		    $dont_remember++;
		}
		if (my $h = $f->{hide_keys}) {
		    if (ref($h) eq "ARRAY") {
			$hide_keys = sub {
			    for my $k (@$h) {
				return 1 if $k eq $_[0];
			    }
			    return 0;
			};
		    }
		}
	    }
	}
	push(@bless, "") if defined($out) && !@bless;
	if (@bless) {
	    $class = shift(@bless);
	    warn "More than one filter callback tried to bless object" if @bless;
	}
    }

    unless ($dont_remember) {
	if (my $s = $seen{$id}) {
	    my($sname, $sidx) = @$s;
	    $refcnt{$sname}++;
	    my $sref = fullname($sname, $sidx,
				($ref && $type eq "SCALAR"));
	    warn "SEEN: [\$$name(@$idx)] => [\$$sname(@$sidx)] ($ref,$sref)" if $DEBUG;
	    return $sref unless $sname eq $name;
	    $refcnt{$name}++;
	    push(@fixup, fullname($name,$idx)." = $sref");
	    return "do{my \$fix}" if @$idx && $idx->[-1] eq '$';
	    return "'fix'";
	}
	$seen{$id} = [$name, $idx];
    }

    if ($class) {
	$pclass = $class;
	$pidx = @$idx;
    }

    if (defined $out) {
	# keep it
    }
    elsif ($type eq "SCALAR" || $type eq "REF" || $type eq "REGEXP") {
	if ($ref) {
	    if ($class && $class eq "Regexp") {
		my $v = "$rval";

		my $mod = "";
		if ($v =~ /^\(\?([msix-]+):([\x00-\xFF]*)\)\z/) {
		    $mod = $1;
		    $v = $2;
		    $mod =~ s/-.*//;
		}

		my $sep = '/';
		my $sep_count = ($v =~ tr/\///);
		if ($sep_count) {
		    # see if we can find a better one
		    for ('|', ',', ':', '#') {
			my $c = eval "\$v =~ tr/\Q$_\E//";
			#print "SEP $_ $c $sep_count\n";
			if ($c < $sep_count) {
			    $sep = $_;
			    $sep_count = $c;
			    last if $sep_count == 0;
			}
		    }
		}
		$v =~ s/\Q$sep\E/\\$sep/g;

		$out = "qr$sep$v$sep$mod";
		undef($class);
	    }
	    else {
		delete $seen{$id} if $type eq "SCALAR";  # will be seen again shortly
		my $val = _dump($$rval, $name, [@$idx, "\$"], 0, $pclass, $pidx);
		$out = $class ? "do{\\(my \$o = $val)}" : "\\$val";
	    }
	} else {
	    if (!defined $$rval) {
		$out = "undef";
	    }
	    elsif ($$rval =~ /^-?[1-9]\d{0,9}\z/ || $$rval eq "0") {
		$out = $$rval;
	    }
	    else {
		$out = str($$rval);
	    }
	    if ($class && !@$idx) {
		# Top is an object, not a reference to one as perl needs
		$refcnt{$name}++;
		my $obj = fullname($name, $idx);
		my $cl  = quote($class);
		push(@fixup, "bless \\$obj, $cl");
	    }
	}
    }
    elsif ($type eq "GLOB") {
	if ($ref) {
	    delete $seen{$id};
	    my $val = _dump($$rval, $name, [@$idx, "*"], 0, $pclass, $pidx);
	    $out = "\\$val";
	    if ($out =~ /^\\\*Symbol::/) {
		$require{Symbol}++;
		$out = "Symbol::gensym()";
	    }
	} else {
	    my $val = "$$rval";
	    $out = "$$rval";

	    for my $k (qw(SCALAR ARRAY HASH)) {
		my $gval = *$$rval{$k};
		next unless defined $gval;
		next if $k eq "SCALAR" && ! defined $$gval;  # always there
		my $f = scalar @fixup;
		push(@fixup, "RESERVED");  # overwritten after _dump() below
		$gval = _dump($gval, $name, [@$idx, "*{$k}"], 0, $pclass, $pidx);
		$refcnt{$name}++;
		my $gname = fullname($name, $idx);
		$fixup[$f] = "$gname = $gval";  #XXX indent $gval
	    }
	}
    }
    elsif ($type eq "ARRAY") {
	my @vals;
	my $tied = tied_str(tied(@$rval));
	my $i = 0;
	for my $v (@$rval) {
	    push(@vals, _dump($v, $name, [@$idx, "[$i]"], $tied, $pclass, $pidx));
	    $i++;
	}
	$out = "[" . format_list(1, $tied, @vals) . "]";
    }
    elsif ($type eq "HASH") {
	my(@keys, @vals);
	my $tied = tied_str(tied(%$rval));

	# statistics to determine variation in key lengths
	my $kstat_max = 0;
	my $kstat_sum = 0;
	my $kstat_sum2 = 0;

	my @orig_keys = keys %$rval;
	if ($hide_keys) {
	    @orig_keys = grep !$hide_keys->($_), @orig_keys;
	}
	my $text_keys = 0;
	for (@orig_keys) {
	    $text_keys++, last unless /^[-+]?(?:0|[1-9]\d*)(?:\.\d+)?\z/;
	}

	if ($text_keys) {
	    @orig_keys = sort { lc($a) cmp lc($b) } @orig_keys;
	}
	else {
	    @orig_keys = sort { $a <=> $b } @orig_keys;
	}

	my $quote;
	for my $key (@orig_keys) {
	    next if $key =~ /^-?[a-zA-Z_]\w*\z/;
	    next if $key =~ /^-?[1-9]\d{0,8}\z/;
	    $quote++;
	    last;
	}

	for my $key (@orig_keys) {
	    my $val = \$rval->{$key};  # capture value before we modify $key
	    $key = quote($key) if $quote;
	    $kstat_max = length($key) if length($key) > $kstat_max;
	    $kstat_sum += length($key);
	    $kstat_sum2 += length($key)*length($key);

	    push(@keys, $key);
	    push(@vals, _dump($$val, $name, [@$idx, "{$key}"], $tied, $pclass, $pidx));
	}
	my $nl = "";
	my $klen_pad = 0;
	my $tmp = "@keys @vals";
	if (length($tmp) > 60 || $tmp =~ /\n/ || $tied) {
	    $nl = "\n";

	    # Determine what padding to add
	    if ($kstat_max < 4) {
		$klen_pad = $kstat_max;
	    }
	    elsif (@keys >= 2) {
		my $n = @keys;
		my $avg = $kstat_sum/$n;
		my $stddev = sqrt(($kstat_sum2 - $n * $avg * $avg) / ($n - 1));

		# I am not actually very happy with this heuristics
		if ($stddev / $kstat_max < 0.25) {
		    $klen_pad = $kstat_max;
		}
		if ($DEBUG) {
		    push(@keys, "__S");
		    push(@vals, sprintf("%.2f (%d/%.1f/%.1f)",
					$stddev / $kstat_max,
					$kstat_max, $avg, $stddev));
		}
	    }
	}
	$out = "{$nl";
	$out .= "$INDENT# $tied$nl" if $tied;
	while (@keys) {
	    my $key = shift @keys;
	    my $val = shift @vals;
	    my $vpad = $INDENT . (" " x ($klen_pad ? $klen_pad + 4 : 0));
	    $val =~ s/\n/\n$vpad/gm;
	    my $kpad = $nl ? $INDENT : " ";
	    $key .= " " x ($klen_pad - length($key)) if $nl;
	    $out .= "$kpad$key => $val,$nl";
	}
	$out =~ s/,$/ / unless $nl;
	$out .= "}";
    }
    elsif ($type eq "CODE") {
	$out = 'sub { ... }';
    }
    else {
	warn "Can't handle $type data";
	$out = "'#$type#'";
    }

    if ($class && $ref) {
	$out = "bless($out, " . quote($class) . ")";
    }
    if ($comment) {
	$comment =~ s/^/# /gm;
	$comment .= "\n" unless $comment =~ /\n\z/;
	$comment =~ s/^#[ \t]+\n/\n/;
	$out = "$comment$out";
    }
    return $out;
}

sub tied_str {
    my $tied = shift;
    if ($tied) {
	if (my $tied_ref = ref($tied)) {
	    $tied = "tied $tied_ref";
	}
	else {
	    $tied = "tied";
	}
    }
    return $tied;
}

sub fullname
{
    my($name, $idx, $ref) = @_;
    substr($name, 0, 0) = "\$";

    my @i = @$idx;  # need copy in order to not modify @$idx
    if ($ref && @i && $i[0] eq "\$") {
	shift(@i);  # remove one deref
	$ref = 0;
    }
    while (@i && $i[0] eq "\$") {
	shift @i;
	$name = "\$$name";
    }

    my $last_was_index;
    for my $i (@i) {
	if ($i eq "*" || $i eq "\$") {
	    $last_was_index = 0;
	    $name = "$i\{$name}";
	} elsif ($i =~ s/^\*//) {
	    $name .= $i;
	    $last_was_index++;
	} else {
	    $name .= "->" unless $last_was_index++;
	    $name .= $i;
	}
    }
    $name = "\\$name" if $ref;
    $name;
}

sub format_list
{
    my $paren = shift;
    my $comment = shift;
    my $indent_lim = $paren ? 0 : 1;
    if (@_ > 3) {
	# can we use range operator to shorten the list?
	my $i = 0;
	while ($i < @_) {
	    my $j = $i + 1;
	    my $v = $_[$i];
	    while ($j < @_) {
		# XXX allow string increment too?
		if ($v eq "0" || $v =~ /^-?[1-9]\d{0,9}\z/) {
		    $v++;
		}
		elsif ($v =~ /^"([A-Za-z]{1,3}\d*)"\z/) {
		    $v = $1;
		    $v++;
		    $v = qq("$v");
		}
		else {
		    last;
		}
		last if $_[$j] ne $v;
		$j++;
	    }
	    if ($j - $i > 3) {
		splice(@_, $i, $j - $i, "$_[$i] .. $_[$j-1]");
	    }
	    $i++;
	}
    }
    my $tmp = "@_";
    if ($comment || (@_ > $indent_lim && (length($tmp) > 60 || $tmp =~ /\n/))) {
	my @elem = @_;
	for (@elem) { s/^/$INDENT/gm; }
	return "\n" . ($comment ? "$INDENT# $comment\n" : "") .
               join(",\n", @elem, "");
    } else {
	return join(", ", @_);
    }
}

sub str {
  if (length($_[0]) > 20) {
      for ($_[0]) {
      # Check for repeated string
      if (/^(.)\1\1\1/s) {
          # seems to be a repating sequence, let's check if it really is
          # without backtracking
          unless (/[^\Q$1\E]/) {
              my $base = quote($1);
              my $repeat = length;
              return "($base x $repeat)"
          }
      }
      # Length protection because the RE engine will blow the stack [RT#33520]
      if (length($_) < 16 * 1024 && /^(.{2,5}?)\1*\z/s) {
	  my $base   = quote($1);
	  my $repeat = length($_)/length($1);
	  return "($base x $repeat)";
      }
      }
  }

  local $_ = &quote;

  if (length($_) > 40  && !/\\x\{/ && length($_) > (length($_[0]) * 2)) {
      # too much binary data, better to represent as a hex/base64 string

      # Base64 is more compact than hex when string is longer than
      # 17 bytes (not counting any require statement needed).
      # But on the other hand, hex is much more readable.
      if ($TRY_BASE64 && length($_[0]) > $TRY_BASE64 &&
	  eval { require MIME::Base64 })
      {
	  $require{"MIME::Base64"}++;
	  return "MIME::Base64::decode(\"" .
	             MIME::Base64::encode($_[0],"") .
		 "\")";
      }
      return "pack(\"H*\",\"" . unpack("H*", $_[0]) . "\")";
  }

  return $_;
}

my %esc = (
    "\a" => "\\a",
    "\b" => "\\b",
    "\t" => "\\t",
    "\n" => "\\n",
    "\f" => "\\f",
    "\r" => "\\r",
    "\e" => "\\e",
);

# put a string value in double quotes
sub quote {
  local($_) = $_[0];
  # If there are many '"' we might want to use qq() instead
  s/([\\\"\@\$])/\\$1/g;
  return qq("$_") unless /[^\040-\176]/;  # fast exit

  s/([\a\b\t\n\f\r\e])/$esc{$1}/g;

  # no need for 3 digits in escape for these
  s/([\0-\037])(?!\d)/sprintf('\\%o',ord($1))/eg;

  s/([\0-\037\177-\377])/sprintf('\\x%02X',ord($1))/eg;
  s/([^\040-\176])/sprintf('\\x{%X}',ord($1))/eg;

  return qq("$_");
}

1;

__END__

=head1 NAME

Data::Dump - Pretty printing of data structures

=head1 SYNOPSIS

 use Data::Dump qw(dump);

 $str = dump(@list);
 @copy_of_list = eval $str;

 # or use it for easy debug printout
 use Data::Dump; dd localtime;

=head1 DESCRIPTION

This module provide a few functions that traverse their
argument and produces a string as its result.  The string contains
Perl code that, when C<eval>ed, produces a deep copy of the original
arguments.

The main feature of the module is that it strives to produce output
that is easy to read.  Example:

    @a = (1, [2, 3], {4 => 5});
    dump(@a);

Produces:

    "(1, [2, 3], { 4 => 5 })"

If you dump just a little data, it is output on a single line. If
you dump data that is more complex or there is a lot of it, line breaks
are automatically added to keep it easy to read.

The following functions are provided (only the dd* functions are exported by default):

=over

=item dump( ... )

=item pp( ... )

Returns a string containing a Perl expression.  If you pass this
string to Perl's built-in eval() function it should return a copy of
the arguments you passed to dump().

If you call the function with multiple arguments then the output will
be wrapped in parenthesis "( ..., ... )".  If you call the function with a
single argument the output will not have the wrapping.  If you call the function with
a single scalar (non-reference) argument it will just return the
scalar quoted if needed, but never break it into multiple lines.  If you
pass multiple arguments or references to arrays of hashes then the
return value might contain line breaks to format it for easier
reading.  The returned string will never be "\n" terminated, even if
contains multiple lines.  This allows code like this to place the
semicolon in the expected place:

   print '$obj = ', dump($obj), ";\n";

If dump() is called in void context, then the dump is printed on
STDERR and then "\n" terminated.  You might find this useful for quick
debug printouts, but the dd*() functions might be better alternatives
for this.

There is no difference between dump() and pp(), except that dump()
shares its name with a not-so-useful perl builtin.  Because of this
some might want to avoid using that name.

=item quote( $string )

Returns a quoted version of the provided string.

It differs from C<dump($string)> in that it will quote even numbers and
not try to come up with clever expressions that might shorten the
output.  If a non-scalar argument is provided then it's just stringified
instead of traversed.

=item dd( ... )

=item ddx( ... )

These functions will call dump() on their argument and print the
result to STDOUT (actually, it's the currently selected output handle, but
STDOUT is the default for that).

The difference between them is only that ddx() will prefix the lines
it prints with "# " and mark the first line with the file and line
number where it was called.  This is meant to be useful for debug
printouts of state within programs.

=item dumpf( ..., \&filter )

Short hand for calling the dump_filtered() function of L<Data::Dump::Filtered>.
This works like dump(), but the last argument should be a filter callback
function.  As objects are visited the filter callback is invoked and it
can modify how the objects are dumped.

=back

=head1 CONFIGURATION

There are a few global variables that can be set to modify the output
generated by the dump functions.  It's wise to localize the setting of
these.

=over

=item $Data::Dump::INDENT

This holds the string that's used for indenting multiline data structures.
It's default value is "  " (two spaces).  Set it to "" to suppress indentation.
Setting it to "| " makes for nice visuals even if the dump output then fails to
be valid Perl.

=item $Data::Dump::TRY_BASE64

How long must a binary string be before we try to use the base64 encoding
for the dump output.  The default is 50.  Set it to 0 to disable base64 dumps.

=back


=head1 LIMITATIONS

Code references will be dumped as C<< sub { ... } >>. Thus, C<eval>ing them will
not reproduce the original routine.  The C<...>-operator used will also require
perl-5.12 or better to be evaled.

If you forget to explicitly import the C<dump> function, your code will
core dump. That's because you just called the builtin C<dump> function
by accident, which intentionally dumps core.  Because of this you can
also import the same function as C<pp>, mnemonic for "pretty-print".

=head1 HISTORY

The C<Data::Dump> module grew out of frustration with Sarathy's
in-most-cases-excellent C<Data::Dumper>.  Basic ideas and some code
are shared with Sarathy's module.

The C<Data::Dump> module provides a much simpler interface than
C<Data::Dumper>.  No OO interface is available and there are fewer
configuration options to worry about.  The other benefit is
that the dump produced does not try to set any variables.  It only
returns what is needed to produce a copy of the arguments.  This means
that C<dump("foo")> simply returns C<'"foo"'>, and C<dump(1..3)> simply
returns C<'(1, 2, 3)'>.

=head1 SEE ALSO

L<Data::Dump::Filtered>, L<Data::Dump::Trace>, L<Data::Dumper>, L<JSON>,
L<Storable>

=head1 AUTHORS

The C<Data::Dump> module is written by Gisle Aas <gisle@aas.no>, based
on C<Data::Dumper> by Gurusamy Sarathy <gsar@umich.edu>.

 Copyright 1998-2010 Gisle Aas.
 Copyright 1996-1998 Gurusamy Sarathy.

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

=cut