[p5sagit/Devel-Size.git] / memnodes.pl

#!/usr/bin/env perl

use strict;
use warnings;
use autodie;

use DBI qw(looks_like_number);
use DBD::SQLite;
use JSON::XS;
use Devel::Dwarn;

use Getopt::Long;

# XXX import these from the XS code
use constant NPtype_NAME     => 0x01;
use constant NPtype_LINK     => 0x02;
use constant NPtype_SV       => 0x03;
use constant NPtype_MAGIC    => 0x04;
use constant NPtype_OP       => 0x05;

use constant NPattr_LEAFSIZE => 0x00;
use constant NPattr_NAME     => 0x01;
use constant NPattr_PADFAKE  => 0x02;
use constant NPattr_PADNAME  => 0x03;
use constant NPattr_PADTMP   => 0x04;
use constant NPattr_NOTE     => 0x05;


GetOptions(
    'text!' => \my $opt_text,
    'dot=s' => \my $opt_dot,
    'db=s'  => \my $opt_db,
    'verbose|v!' => \my $opt_verbose,
    'debug|d!' => \my $opt_debug,
    'showid!' => \my $opt_showid,
) or exit 1;

my $j = JSON::XS->new->ascii->pretty(0);

my ($dbh, $node_ins_sth);
if ($opt_db) {
    $dbh = DBI->connect("dbi:SQLite:dbname=$opt_db","","", {
        RaiseError => 1, PrintError => 0, AutoCommit => 0
    });
    $dbh->do("PRAGMA synchronous = OFF");
    $dbh->do("DROP TABLE IF EXISTS node");
    $dbh->do(q{
        CREATE TABLE node (
            id integer primary key,
            name text,
            title text,
            type integer,
            depth integer,
            parent_id integer,

            self_size integer,
            kids_size integer,
            kids_node_count integer,
            child_ids text,
            attr_json text,
            leaves_json text
        )
    });
    $node_ins_sth = $dbh->prepare(q{
        INSERT INTO node VALUES (?,?,?,?,?,?,  ?,?,?,?,?,?)
    });
}

my @stack;
my %seqn2node;

use HTML::Entities qw(encode_entities);;
my $dotnode = sub {
    my $name = encode_entities(shift);
    $name =~ s/"/\\"/g;
    return '"'.$name.'"';
};


my $dot_fh;
if ($opt_dot) {
    open $dot_fh, ">$opt_dot";
    print $dot_fh "digraph {\n"; # }
    print $dot_fh "graph [overlap=false]\n"; # target="???", URL="???"
}

sub fmt_size {
    my $size = shift;
    my $kb = $size / 1024;
    return $size if $kb < 5;
    return sprintf "%.1fKb", $kb if $kb < 1000;
    return sprintf "%.1fMb", $kb/1024;
}


sub enter_node {
    my $x = shift;

    my $parent = $stack[-1];
    if ($parent) {

        # If node is an AVelem of a CvPADLIST propagate pad name to AVelem
        if ($parent->{name} eq 'SV(PVAV)' && @stack >= 4) {
            if ((my $cvpl = $stack[-4])->{name} eq 'CvPADLIST') {
                my $padnames = $cvpl->{attr}{+NPattr_PADNAME} || [];
                warn "@$padnames";
                $x->{name} = "pad";
            }
        }
    }

    return $x;
}


sub leave_node {
    my $x = shift;
    delete $seqn2node{$x->{id}};

    my $self_size = 0; $self_size += $_  for values %{$x->{leaves}};
    $x->{self_size} = $self_size;

    my $parent = $stack[-1];
    if ($parent) {
        # link to parent
        $x->{parent_id} = $parent->{id};
        # accumulate into parent
        $parent->{kids_node_count} += 1 + ($x->{kids_node_count}||0);
        $parent->{kids_size} += $self_size + $x->{kids_size};
        push @{$parent->{child_id}}, $x->{id};
    }

    # output
    # ...
    if ($opt_dot) {
        printf "// n%d parent=%s(type=%s)\n", $x->{id},
                $parent ? $parent->{id} : "",
                $parent ? $parent->{type} : ""
            if 0;
        if ($x->{type} != NPtype_LINK) {
            my $name = $x->{title} ? "\"$x->{title}\" $x->{name}" : $x->{name};

            if ($x->{kids_size}) {
                $name .= sprintf " %s+%s=%s", fmt_size($x->{self_size}), fmt_size($x->{kids_size}), fmt_size($x->{self_size}+$x->{kids_size});
            }
            else {
                $name .= sprintf " +%s", fmt_size($x->{self_size});
            }
            $name .= " $x->{id}" if $opt_showid;

            my @node_attr = (
                sprintf("label=%s", $dotnode->($name)),
                "id=$x->{id}",
            );
            my @link_attr;
            #if ($x->{name} eq 'hek') { push @node_attr, "shape=point"; push @node_attr, "labelfontsize=6"; }
            if ($parent) { # probably a link
                my $parent_id = $parent->{id};
                my @link_attr = ("id=$parent_id");
                if ($parent->{type} == NPtype_LINK) { # link
                    (my $link_name = $parent->{name}) =~ s/->$//;
                    push @link_attr, (sprintf "label=%s", $dotnode->($link_name));
                    $parent_id = ($stack[-2]||die "panic")->{id};
                }
                printf $dot_fh qq{n%d -> n%d [%s];\n},
                    $parent_id, $x->{id}, join(",", @link_attr);
            }
            printf $dot_fh qq{n%d [ %s ];\n}, $x->{id}, join(",", @node_attr);
        }

    }
    if ($dbh) {
        my $attr_json = $j->encode($x->{attr});
        my $leaves_json = $j->encode($x->{leaves});
        $node_ins_sth->execute(
            $x->{id}, $x->{name}, $x->{title}, $x->{type}, $x->{depth}, $x->{parent_id},
            $x->{self_size}, $x->{kids_size}, $x->{kids_node_count},
            $x->{child_id} ? join(",", @{$x->{child_id}}) : undef,
            $attr_json, $leaves_json,
        );
        # XXX attribs
    }
    return;
}

my $indent = ":   ";
my @attr_type_name = (qw(size NAME PADFAKE my PADTMP NOTE));

while (<>) {
    chomp;

    my ($type, $id, $val, $name, $extra) = split / /, $_, 5;

    if ($type =~ s/^-//) {     # Node type ($val is depth)
        printf "%s%s %s [#%d @%d]\n", $indent x $val, $name, $extra||'', $id, $val
            if $opt_text;
        while ($val < @stack) {
            leave_node(my $x = pop @stack);
            warn "N $id d$val ends $x->{id} d$x->{depth}: size $x->{self_size}+$x->{kids_size}\n"
                if $opt_verbose;
        }
        die "panic: stack already has item at depth $val"
            if $stack[$val];
        my $node = enter_node({
            id => $id, type => $type, name => $name, extra => $extra,
            attr => {}, leaves => {}, depth => $val, self_size=>0, kids_size=>0
        });
        $stack[$val] = $node;
        $seqn2node{$id} = $node;
    }
    # --- Leaf name and memory size
    elsif ($type eq "L") {
        my $node = $seqn2node{$id} || die;
        $node->{leaves}{$name} += $val;
        printf "%s+%d %s\n", $indent x ($node->{depth}+1), $val, $name
            if $opt_text;
    }
    # --- Attribute type, name and value
    elsif (looks_like_number($type)) {
        my $node = $seqn2node{$id} || die;
        my $attr = $node->{attr} || die;

        if ($type == NPattr_NAME or $type == NPattr_NOTE) {
            printf "%s~%s(%s) %d [t%d]\n", $indent x ($node->{depth}+1), $attr_type_name[$type], $name, $val, $type
                if $opt_text;
            warn "Node $id already has attribute $type:$name (value $attr->{$type}{$name})\n"
                if exists $attr->{$type}{$name};
            $attr->{$type}{$name} = $val || $id;
            $node->{title} = $name if $type == 1 and !$val;
        }
        elsif (NPattr_PADFAKE==$type or NPattr_PADTMP==$type or NPattr_PADNAME==$type) {
            printf "%s~%s('%s') %d [t%d]\n", $indent x ($node->{depth}+1), $attr_type_name[$type], $name, $val, $type
                if $opt_text;
            warn "Node $id already has attribute $type:$name (value $attr->{$type}[$val])\n"
                if defined $attr->{$type}[$val];
            $attr->{+NPattr_PADNAME}[$val] = $name; # store all as NPattr_PADNAME
            Dwarn $attr;
        }
        else {
            printf "%s~%s %d [t%d]\n", $indent x ($node->{depth}+1), $name, $val, $type
                if $opt_text;
            warn "Invalid attribute type '$type' on line $. ($_)";
        }
    }
    else {
        warn "Invalid type '$type' on line $. ($_)";
        next;
    }
    $dbh->commit if $dbh and $id % 10_000 == 0;
}

my $top = $stack[0]; # grab top node before we pop all the nodes
leave_node(pop @stack) while @stack;
warn "EOF ends $top->{id} d$top->{depth}: size $top->{self_size}+$top->{kids_size}\n"
    if $opt_verbose;
warn Dumper($top) if $opt_verbose;

if ($dot_fh) {
    print $dot_fh "}\n";
    close $dot_fh;
    system("open -a Graphviz $opt_dot");
}

$dbh->commit if $dbh;

use Data::Dumper;
warn Dumper(\%seqn2node) if %seqn2node; # should be empty

=for
SV(PVAV) fill=1/1       [#1 @0] 
:   +64 sv =64 
:   +16 av_max =80 
:   AVelem->        [#2 @1] 
:   :   SV(RV)      [#3 @2] 
:   :   :   +24 sv =104 
:   :   :   RV->        [#4 @3] 
:   :   :   :   SV(PVAV) fill=-1/-1     [#5 @4] 
:   :   :   :   :   +64 sv =168 
:   AVelem->        [#6 @1] 
:   :   SV(IV)      [#7 @2] 
:   :   :   +24 sv =192 
192 at -e line 1.
=cut
__DATA__
N 1 0 SV(PVAV) fill=1/1
L 1 64 sv
L 1 16 av_max
N 2 1 AVelem->
N 3 2 SV(RV)
L 3 24 sv
N 4 3 RV->
N 5 4 SV(PVAV) fill=-1/-1
L 5 64 sv
N 6 1 AVelem->
N 7 2 SV(IV)
L 7 24 sv
Commit	Line	Data
5aa3ad8e	1	#!/usr/bin/env perl
2c631ee0	2
	3	use strict;
	4	use warnings;
5aa3ad8e	5	use autodie;
2c631ee0	6
e8f4c506	7	use DBI qw(looks_like_number);
b2fc39a5	8	use DBD::SQLite;
f60f09e5	9	use JSON::XS;
de73b186	10	use Devel::Dwarn;
b2fc39a5	11
fc6614ee	12	use Getopt::Long;
fc6614ee	13
de73b186	14	# XXX import these from the XS code
	15	use constant NPtype_NAME => 0x01;
	16	use constant NPtype_LINK => 0x02;
	17	use constant NPtype_SV => 0x03;
	18	use constant NPtype_MAGIC => 0x04;
	19	use constant NPtype_OP => 0x05;
	20
	21	use constant NPattr_LEAFSIZE => 0x00;
	22	use constant NPattr_NAME => 0x01;
	23	use constant NPattr_PADFAKE => 0x02;
	24	use constant NPattr_PADNAME => 0x03;
	25	use constant NPattr_PADTMP => 0x04;
	26	use constant NPattr_NOTE => 0x05;
	27
	28
fc6614ee	29	GetOptions(
5aa3ad8e	30	'text!' => \my $opt_text,
5aa3ad8e	31	'dot=s' => \my $opt_dot,
b2fc39a5	32	'db=s' => \my $opt_db,
e8f4c506	33	'verbose\|v!' => \my $opt_verbose,
e8f4c506	34	'debug\|d!' => \my $opt_debug,
98128850	35	'showid!' => \my $opt_showid,
fc6614ee	36	) or exit 1;
94fab3d1	37
f60f09e5	38	my $j = JSON::XS->new->ascii->pretty(0);
f60f09e5	39
1915946b	40	my ($dbh, $node_ins_sth);
	41	if ($opt_db) {
	42	$dbh = DBI->connect("dbi:SQLite:dbname=$opt_db","","", {
	43	RaiseError => 1, PrintError => 0, AutoCommit => 0
	44	});
	45	$dbh->do("PRAGMA synchronous = OFF");
	46	$dbh->do("DROP TABLE IF EXISTS node");
	47	$dbh->do(q{
	48	CREATE TABLE node (
	49	id integer primary key,
	50	name text,
	51	title text,
98128850	52	type integer,
1915946b	53	depth integer,
	54	parent_id integer,
	55
	56	self_size integer,
	57	kids_size integer,
	58	kids_node_count integer,
	59	child_ids text,
	60	attr_json text,
	61	leaves_json text
	62	)
	63	});
	64	$node_ins_sth = $dbh->prepare(q{
98128850	65	INSERT INTO node VALUES (?,?,?,?,?,?, ?,?,?,?,?,?)
1915946b	66	});
1915946b	67	}
b2fc39a5	68
2c631ee0	69	my @stack;
	70	my %seqn2node;
	71
7020702a	72	use HTML::Entities qw(encode_entities);;
	73	my $dotnode = sub {
	74	my $name = encode_entities(shift);
	75	$name =~ s/"/\\"/g;
7020702a	76	return '"'.$name.'"';
7020702a	77	};
ee2793c1	78
ee2793c1	79
5aa3ad8e	80	my $dot_fh;
ee2793c1	81	if ($opt_dot) {
5aa3ad8e	82	open $dot_fh, ">$opt_dot";
	83	print $dot_fh "digraph {\n"; # }
	84	print $dot_fh "graph [overlap=false]\n"; # target="???", URL="???"
ee2793c1	85	}
ee2793c1	86
0741448c	87	sub fmt_size {
	88	my $size = shift;
	89	my $kb = $size / 1024;
	90	return $size if $kb < 5;
	91	return sprintf "%.1fKb", $kb if $kb < 1000;
	92	return sprintf "%.1fMb", $kb/1024;
	93	}
	94
ee2793c1	95
94fab3d1	96	sub enter_node {
94fab3d1	97	my $x = shift;
de73b186	98
	99	my $parent = $stack[-1];
	100	if ($parent) {
	101
	102	# If node is an AVelem of a CvPADLIST propagate pad name to AVelem
	103	if ($parent->{name} eq 'SV(PVAV)' && @stack >= 4) {
	104	if ((my $cvpl = $stack[-4])->{name} eq 'CvPADLIST') {
	105	my $padnames = $cvpl->{attr}{+NPattr_PADNAME} \|\| [];
	106	warn "@$padnames";
	107	$x->{name} = "pad";
	108	}
	109	}
ee2793c1	110	}
de73b186	111
de73b186	112	return $x;
94fab3d1	113	}
94fab3d1	114
de73b186	115
94fab3d1	116	sub leave_node {
94fab3d1	117	my $x = shift;
b2fc39a5	118	delete $seqn2node{$x->{id}};
ee2793c1	119
94fab3d1	120	my $self_size = 0; $self_size += $_ for values %{$x->{leaves}};
94fab3d1	121	$x->{self_size} = $self_size;
ee2793c1	122
	123	my $parent = $stack[-1];
	124	if ($parent) {
2c631ee0	125	# link to parent
5a78486c	126	$x->{parent_id} = $parent->{id};
2c631ee0	127	# accumulate into parent
2c631ee0	128	$parent->{kids_node_count} += 1 + ($x->{kids_node_count}\|\|0);
94fab3d1	129	$parent->{kids_size} += $self_size + $x->{kids_size};
5a78486c	130	push @{$parent->{child_id}}, $x->{id};
2c631ee0	131	}
de73b186	132
2c631ee0	133	# output
2c631ee0	134	# ...
ee2793c1	135	if ($opt_dot) {
1915946b	136	printf "// n%d parent=%s(type=%s)\n", $x->{id},
0741448c	137	$parent ? $parent->{id} : "",
	138	$parent ? $parent->{type} : ""
	139	if 0;
de73b186	140	if ($x->{type} != NPtype_LINK) {
0741448c	141	my $name = $x->{title} ? "\"$x->{title}\" $x->{name}" : $x->{name};
	142
	143	if ($x->{kids_size}) {
	144	$name .= sprintf " %s+%s=%s", fmt_size($x->{self_size}), fmt_size($x->{kids_size}), fmt_size($x->{self_size}+$x->{kids_size});
	145	}
	146	else {
	147	$name .= sprintf " +%s", fmt_size($x->{self_size});
	148	}
98128850	149	$name .= " $x->{id}" if $opt_showid;
0741448c	150
	151	my @node_attr = (
	152	sprintf("label=%s", $dotnode->($name)),
	153	"id=$x->{id}",
	154	);
	155	my @link_attr;
	156	#if ($x->{name} eq 'hek') { push @node_attr, "shape=point"; push @node_attr, "labelfontsize=6"; }
1915946b	157	if ($parent) { # probably a link
1915946b	158	my $parent_id = $parent->{id};
0741448c	159	my @link_attr = ("id=$parent_id");
de73b186	160	if ($parent->{type} == NPtype_LINK) { # link
1915946b	161	(my $link_name = $parent->{name}) =~ s/->$//;
	162	push @link_attr, (sprintf "label=%s", $dotnode->($link_name));
	163	$parent_id = ($stack[-2]\|\|die "panic")->{id};
	164	}
5aa3ad8e	165	printf $dot_fh qq{n%d -> n%d [%s];\n},
1915946b	166	$parent_id, $x->{id}, join(",", @link_attr);
1915946b	167	}
5aa3ad8e	168	printf $dot_fh qq{n%d [ %s ];\n}, $x->{id}, join(",", @node_attr);
1915946b	169	}
1915946b	170
ee2793c1	171	}
b2fc39a5	172	if ($dbh) {
f60f09e5	173	my $attr_json = $j->encode($x->{attr});
e78b28ca	174	my $leaves_json = $j->encode($x->{leaves});
b2fc39a5	175	$node_ins_sth->execute(
98128850	176	$x->{id}, $x->{name}, $x->{title}, $x->{type}, $x->{depth}, $x->{parent_id},
b2fc39a5	177	$x->{self_size}, $x->{kids_size}, $x->{kids_node_count},
f60f09e5	178	$x->{child_id} ? join(",", @{$x->{child_id}}) : undef,
e78b28ca	179	$attr_json, $leaves_json,
b2fc39a5	180	);
	181	# XXX attribs
	182	}
94fab3d1	183	return;
2c631ee0	184	}
2c631ee0	185
5aa3ad8e	186	my $indent = ": ";
de73b186	187	my @attr_type_name = (qw(size NAME PADFAKE my PADTMP NOTE));
94fab3d1	188
2c631ee0	189	while (<>) {
2c631ee0	190	chomp;
de73b186	191
b2fc39a5	192	my ($type, $id, $val, $name, $extra) = split / /, $_, 5;
de73b186	193
ee2793c1	194	if ($type =~ s/^-//) { # Node type ($val is depth)
5aa3ad8e	195	printf "%s%s %s [#%d @%d]\n", $indent x $val, $name, $extra\|\|'', $id, $val
5aa3ad8e	196	if $opt_text;
2c631ee0	197	while ($val < @stack) {
94fab3d1	198	leave_node(my $x = pop @stack);
e8f4c506	199	warn "N $id d$val ends $x->{id} d$x->{depth}: size $x->{self_size}+$x->{kids_size}\n"
e8f4c506	200	if $opt_verbose;
2c631ee0	201	}
c5078bcb	202	die "panic: stack already has item at depth $val"
c5078bcb	203	if $stack[$val];
de73b186	204	my $node = enter_node({
	205	id => $id, type => $type, name => $name, extra => $extra,
	206	attr => {}, leaves => {}, depth => $val, self_size=>0, kids_size=>0
	207	});
	208	$stack[$val] = $node;
b2fc39a5	209	$seqn2node{$id} = $node;
2c631ee0	210	}
de73b186	211	# --- Leaf name and memory size
de73b186	212	elsif ($type eq "L") {
b2fc39a5	213	my $node = $seqn2node{$id} \|\| die;
2c631ee0	214	$node->{leaves}{$name} += $val;
5aa3ad8e	215	printf "%s+%d %s\n", $indent x ($node->{depth}+1), $val, $name
5aa3ad8e	216	if $opt_text;
2c631ee0	217	}
de73b186	218	# --- Attribute type, name and value
de73b186	219	elsif (looks_like_number($type)) {
b2fc39a5	220	my $node = $seqn2node{$id} \|\| die;
e8f4c506	221	my $attr = $node->{attr} \|\| die;
de73b186	222
	223	if ($type == NPattr_NAME or $type == NPattr_NOTE) {
	224	printf "%s~%s(%s) %d [t%d]\n", $indent x ($node->{depth}+1), $attr_type_name[$type], $name, $val, $type
	225	if $opt_text;
e8f4c506	226	warn "Node $id already has attribute $type:$name (value $attr->{$type}{$name})\n"
	227	if exists $attr->{$type}{$name};
	228	$attr->{$type}{$name} = $val \|\| $id;
e8f4c506	229	$node->{title} = $name if $type == 1 and !$val;
e8f4c506	230	}
de73b186	231	elsif (NPattr_PADFAKE==$type or NPattr_PADTMP==$type or NPattr_PADNAME==$type) {
	232	printf "%s~%s('%s') %d [t%d]\n", $indent x ($node->{depth}+1), $attr_type_name[$type], $name, $val, $type
	233	if $opt_text;
e8f4c506	234	warn "Node $id already has attribute $type:$name (value $attr->{$type}[$val])\n"
e8f4c506	235	if defined $attr->{$type}[$val];
de73b186	236	$attr->{+NPattr_PADNAME}[$val] = $name; # store all as NPattr_PADNAME
de73b186	237	Dwarn $attr;
e8f4c506	238	}
e8f4c506	239	else {
de73b186	240	printf "%s~%s %d [t%d]\n", $indent x ($node->{depth}+1), $name, $val, $type
de73b186	241	if $opt_text;
e8f4c506	242	warn "Invalid attribute type '$type' on line $. ($_)";
e8f4c506	243	}
2c631ee0	244	}
	245	else {
	246	warn "Invalid type '$type' on line $. ($_)";
e8f4c506	247	next;
2c631ee0	248	}
b2fc39a5	249	$dbh->commit if $dbh and $id % 10_000 == 0;
2c631ee0	250	}
2c631ee0	251
c5078bcb	252	my $top = $stack[0]; # grab top node before we pop all the nodes
	253	leave_node(pop @stack) while @stack;
	254	warn "EOF ends $top->{id} d$top->{depth}: size $top->{self_size}+$top->{kids_size}\n"
	255	if $opt_verbose;
	256	warn Dumper($top) if $opt_verbose;
5aa3ad8e	257
	258	if ($dot_fh) {
	259	print $dot_fh "}\n";
	260	close $dot_fh;
	261	system("open -a Graphviz $opt_dot");
2c631ee0	262	}
94fab3d1	263
b2fc39a5	264	$dbh->commit if $dbh;
b2fc39a5	265
2c631ee0	266	use Data::Dumper;
5aa3ad8e	267	warn Dumper(\%seqn2node) if %seqn2node; # should be empty
2c631ee0	268
	269	=for
	270	SV(PVAV) fill=1/1 [#1 @0]
	271	: +64 sv =64
	272	: +16 av_max =80
	273	: AVelem-> [#2 @1]
	274	: : SV(RV) [#3 @2]
	275	: : : +24 sv =104
	276	: : : RV-> [#4 @3]
	277	: : : : SV(PVAV) fill=-1/-1 [#5 @4]
	278	: : : : : +64 sv =168
	279	: AVelem-> [#6 @1]
	280	: : SV(IV) [#7 @2]
	281	: : : +24 sv =192
	282	192 at -e line 1.
	283	=cut
	284	__DATA__
	285	N 1 0 SV(PVAV) fill=1/1
	286	L 1 64 sv
	287	L 1 16 av_max
	288	N 2 1 AVelem->
	289	N 3 2 SV(RV)
	290	L 3 24 sv
	291	N 4 3 RV->
	292	N 5 4 SV(PVAV) fill=-1/-1
	293	L 5 64 sv
	294	N 6 1 AVelem->
	295	N 7 2 SV(IV)
	296	L 7 24 sv