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

#!/bin/env perl

use strict;
use warnings;

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

use Getopt::Long;

GetOptions(
    'json!' => \my $opt_json,
    'dot!' => \my $opt_dot,
    'db=s'  => \my $opt_db,
    'verbose|v!' => \my $opt_verbose,
    'debug|d!' => \my $opt_debug,
) 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,
            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.'"';
};

print "memnodes = [" if $opt_json;

if ($opt_dot) {
    print "digraph {\n"; # }
    print "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 $x = shift;
    if ($opt_json) {
        print "    " x $x->{depth};
        print qq({ "id": "$x->{id}", "name": "$x->{name}", "depth":$x->{depth}, "children":[ \n);
    }
    if ($opt_dot) {
        #printf $fh qq{\tn%d [ %s ]\n}, $x->{id}, $dotnode->($x->{name});
        #print qq({ "id": "$x->{id}", "name": "$x->{name}", "depth":$x->{depth}, "children":[ \n);
    }
    return;
}

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_json) {
        print "    " x $x->{depth};
        my $size = $self_size + $x->{kids_size};
        print qq(], "data":{ "\$area": $size } },\n);
    }
    if ($opt_dot) {
        printf "// n%d parent=%s(type=%s)\n", $x->{id},
                $parent ? $parent->{id} : "",
                $parent ? $parent->{type} : ""
            if 0;
        if ($x->{type} != 2) {
            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});
            }

            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} == 2) { # link
                    (my $link_name = $parent->{name}) =~ s/->$//;
                    push @link_attr, (sprintf "label=%s", $dotnode->($link_name));
                    $parent_id = ($stack[-2]||die "panic")->{id};
                }
                printf qq{n%d -> n%d [%s];\n},
                    $parent_id, $x->{id}, join(",", @link_attr);
            }
            printf 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->{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;
}


while (<>) {
    chomp;
    my ($type, $id, $val, $name, $extra) = split / /, $_, 5;
    if ($type =~ s/^-//) {     # Node type ($val is depth)
        while ($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 1 if $stack[$val];
        my $node = $stack[$val] = { id => $id, type => $type, name => $name, extra => $extra, attr => {}, leaves => {}, depth => $val, self_size=>0, kids_size=>0 };
        enter_node($node);
        $seqn2node{$id} = $node;
    }
    elsif ($type eq "L") {  # Leaf name and memory size
        my $node = $seqn2node{$id} || die;
        $node->{leaves}{$name} += $val;
    }
    elsif (looks_like_number($type)) {  # Attribute type, name and value
        my $node = $seqn2node{$id} || die;
        my $attr = $node->{attr} || die;
        if ($type == 1) { # NPattr_NAME
            warn "Node $id already has attribute $type:$name (value $attr->{$type}{$name})\n"
                if exists $attr->{$type}{$name};
            $attr->{$type}{$name} = $val || $id;
            warn "A \@$id: '$name' $val\n";
            $node->{title} = $name if $type == 1 and !$val;
        }
        elsif (2 <= $type and $type <= 4) { # NPattr_PAD*
            warn "Node $id already has attribute $type:$name (value $attr->{$type}[$val])\n"
                if defined $attr->{$type}[$val];
            $attr->{$type}[$val] = $name;
        }
        else {
            warn "Invalid attribute type '$type' on line $. ($_)";
        }
    }
    else {
        warn "Invalid type '$type' on line $. ($_)";
        next;
    }
    $dbh->commit if $dbh and $id % 10_000 == 0;
}

my $x;
while (@stack > 1) {
    leave_node($x = pop @stack) while @stack;
    warn "EOF ends $x->{id} d$x->{depth}: size $x->{self_size}+$x->{kids_size}\n";
}
print " ];\n" if $opt_json;
print "}\n" if $opt_dot;

$dbh->commit if $dbh;

use Data::Dumper;
warn Dumper(\$x);
warn Dumper(\%seqn2node);

=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
2c631ee0	1	#!/bin/env perl
	2
	3	use strict;
	4	use warnings;
	5
e8f4c506	6	use DBI qw(looks_like_number);
b2fc39a5	7	use DBD::SQLite;
f60f09e5	8	use JSON::XS;
b2fc39a5	9
fc6614ee	10	use Getopt::Long;
	11
	12	GetOptions(
	13	'json!' => \my $opt_json,
ee2793c1	14	'dot!' => \my $opt_dot,
b2fc39a5	15	'db=s' => \my $opt_db,
e8f4c506	16	'verbose\|v!' => \my $opt_verbose,
e8f4c506	17	'debug\|d!' => \my $opt_debug,
fc6614ee	18	) or exit 1;
94fab3d1	19
f60f09e5	20	my $j = JSON::XS->new->ascii->pretty(0);
f60f09e5	21
1915946b	22	my ($dbh, $node_ins_sth);
	23	if ($opt_db) {
	24	$dbh = DBI->connect("dbi:SQLite:dbname=$opt_db","","", {
	25	RaiseError => 1, PrintError => 0, AutoCommit => 0
	26	});
	27	$dbh->do("PRAGMA synchronous = OFF");
	28	$dbh->do("DROP TABLE IF EXISTS node");
	29	$dbh->do(q{
	30	CREATE TABLE node (
	31	id integer primary key,
	32	name text,
	33	title text,
	34	depth integer,
	35	parent_id integer,
	36
	37	self_size integer,
	38	kids_size integer,
	39	kids_node_count integer,
	40	child_ids text,
	41	attr_json text,
	42	leaves_json text
	43	)
	44	});
	45	$node_ins_sth = $dbh->prepare(q{
	46	INSERT INTO node VALUES (?,?,?,?,?, ?,?,?,?,?,?)
	47	});
	48	}
b2fc39a5	49
2c631ee0	50	my @stack;
	51	my %seqn2node;
	52
7020702a	53	use HTML::Entities qw(encode_entities);;
	54	my $dotnode = sub {
	55	my $name = encode_entities(shift);
	56	$name =~ s/"/\\"/g;
7020702a	57	return '"'.$name.'"';
7020702a	58	};
ee2793c1	59
	60	print "memnodes = [" if $opt_json;
	61
	62	if ($opt_dot) {
	63	print "digraph {\n"; # }
	64	print "graph [overlap=false]\n"; # target="???", URL="???"
	65	}
	66
0741448c	67	sub fmt_size {
	68	my $size = shift;
	69	my $kb = $size / 1024;
	70	return $size if $kb < 5;
	71	return sprintf "%.1fKb", $kb if $kb < 1000;
	72	return sprintf "%.1fMb", $kb/1024;
	73	}
	74
ee2793c1	75
94fab3d1	76	sub enter_node {
	77	my $x = shift;
	78	if ($opt_json) {
	79	print " " x $x->{depth};
b2fc39a5	80	print qq({ "id": "$x->{id}", "name": "$x->{name}", "depth":$x->{depth}, "children":[ \n);
94fab3d1	81	}
ee2793c1	82	if ($opt_dot) {
	83	#printf $fh qq{\tn%d [ %s ]\n}, $x->{id}, $dotnode->($x->{name});
	84	#print qq({ "id": "$x->{id}", "name": "$x->{name}", "depth":$x->{depth}, "children":[ \n);
	85	}
94fab3d1	86	return;
	87	}
	88
	89	sub leave_node {
	90	my $x = shift;
b2fc39a5	91	delete $seqn2node{$x->{id}};
ee2793c1	92
94fab3d1	93	my $self_size = 0; $self_size += $_ for values %{$x->{leaves}};
94fab3d1	94	$x->{self_size} = $self_size;
ee2793c1	95
	96	my $parent = $stack[-1];
	97	if ($parent) {
2c631ee0	98	# link to parent
5a78486c	99	$x->{parent_id} = $parent->{id};
2c631ee0	100	# accumulate into parent
2c631ee0	101	$parent->{kids_node_count} += 1 + ($x->{kids_node_count}\|\|0);
94fab3d1	102	$parent->{kids_size} += $self_size + $x->{kids_size};
5a78486c	103	push @{$parent->{child_id}}, $x->{id};
2c631ee0	104	}
	105	# output
	106	# ...
94fab3d1	107	if ($opt_json) {
	108	print " " x $x->{depth};
	109	my $size = $self_size + $x->{kids_size};
	110	print qq(], "data":{ "\$area": $size } },\n);
	111	}
ee2793c1	112	if ($opt_dot) {
1915946b	113	printf "// n%d parent=%s(type=%s)\n", $x->{id},
0741448c	114	$parent ? $parent->{id} : "",
	115	$parent ? $parent->{type} : ""
	116	if 0;
1915946b	117	if ($x->{type} != 2) {
0741448c	118	my $name = $x->{title} ? "\"$x->{title}\" $x->{name}" : $x->{name};
	119
	120	if ($x->{kids_size}) {
	121	$name .= sprintf " %s+%s=%s", fmt_size($x->{self_size}), fmt_size($x->{kids_size}), fmt_size($x->{self_size}+$x->{kids_size});
	122	}
	123	else {
	124	$name .= sprintf " +%s", fmt_size($x->{self_size});
	125	}
	126
	127	my @node_attr = (
	128	sprintf("label=%s", $dotnode->($name)),
	129	"id=$x->{id}",
	130	);
	131	my @link_attr;
	132	#if ($x->{name} eq 'hek') { push @node_attr, "shape=point"; push @node_attr, "labelfontsize=6"; }
1915946b	133	if ($parent) { # probably a link
1915946b	134	my $parent_id = $parent->{id};
0741448c	135	my @link_attr = ("id=$parent_id");
1915946b	136	if ($parent->{type} == 2) { # link
	137	(my $link_name = $parent->{name}) =~ s/->$//;
	138	push @link_attr, (sprintf "label=%s", $dotnode->($link_name));
	139	$parent_id = ($stack[-2]\|\|die "panic")->{id};
	140	}
	141	printf qq{n%d -> n%d [%s];\n},
	142	$parent_id, $x->{id}, join(",", @link_attr);
	143	}
	144	printf qq{n%d [ %s ];\n}, $x->{id}, join(",", @node_attr);
	145	}
	146
ee2793c1	147	}
b2fc39a5	148	if ($dbh) {
f60f09e5	149	my $attr_json = $j->encode($x->{attr});
e78b28ca	150	my $leaves_json = $j->encode($x->{leaves});
b2fc39a5	151	$node_ins_sth->execute(
e8f4c506	152	$x->{id}, $x->{name}, $x->{title}, $x->{depth}, $x->{parent_id},
b2fc39a5	153	$x->{self_size}, $x->{kids_size}, $x->{kids_node_count},
f60f09e5	154	$x->{child_id} ? join(",", @{$x->{child_id}}) : undef,
e78b28ca	155	$attr_json, $leaves_json,
b2fc39a5	156	);
	157	# XXX attribs
	158	}
94fab3d1	159	return;
2c631ee0	160	}
2c631ee0	161
94fab3d1	162
2c631ee0	163	while (<>) {
2c631ee0	164	chomp;
b2fc39a5	165	my ($type, $id, $val, $name, $extra) = split / /, $_, 5;
ee2793c1	166	if ($type =~ s/^-//) { # Node type ($val is depth)
2c631ee0	167	while ($val < @stack) {
94fab3d1	168	leave_node(my $x = pop @stack);
e8f4c506	169	warn "N $id d$val ends $x->{id} d$x->{depth}: size $x->{self_size}+$x->{kids_size}\n"
e8f4c506	170	if $opt_verbose;
2c631ee0	171	}
2c631ee0	172	die 1 if $stack[$val];
ee2793c1	173	my $node = $stack[$val] = { id => $id, type => $type, name => $name, extra => $extra, attr => {}, leaves => {}, depth => $val, self_size=>0, kids_size=>0 };
94fab3d1	174	enter_node($node);
b2fc39a5	175	$seqn2node{$id} = $node;
2c631ee0	176	}
2c631ee0	177	elsif ($type eq "L") { # Leaf name and memory size
b2fc39a5	178	my $node = $seqn2node{$id} \|\| die;
2c631ee0	179	$node->{leaves}{$name} += $val;
2c631ee0	180	}
e8f4c506	181	elsif (looks_like_number($type)) { # Attribute type, name and value
b2fc39a5	182	my $node = $seqn2node{$id} \|\| die;
e8f4c506	183	my $attr = $node->{attr} \|\| die;
	184	if ($type == 1) { # NPattr_NAME
	185	warn "Node $id already has attribute $type:$name (value $attr->{$type}{$name})\n"
	186	if exists $attr->{$type}{$name};
	187	$attr->{$type}{$name} = $val \|\| $id;
	188	warn "A \@$id: '$name' $val\n";
	189	$node->{title} = $name if $type == 1 and !$val;
	190	}
	191	elsif (2 <= $type and $type <= 4) { # NPattr_PAD*
	192	warn "Node $id already has attribute $type:$name (value $attr->{$type}[$val])\n"
	193	if defined $attr->{$type}[$val];
	194	$attr->{$type}[$val] = $name;
	195	}
	196	else {
	197	warn "Invalid attribute type '$type' on line $. ($_)";
	198	}
2c631ee0	199	}
	200	else {
	201	warn "Invalid type '$type' on line $. ($_)";
e8f4c506	202	next;
2c631ee0	203	}
b2fc39a5	204	$dbh->commit if $dbh and $id % 10_000 == 0;
2c631ee0	205	}
	206
	207	my $x;
	208	while (@stack > 1) {
94fab3d1	209	leave_node($x = pop @stack) while @stack;
b2fc39a5	210	warn "EOF ends $x->{id} d$x->{depth}: size $x->{self_size}+$x->{kids_size}\n";
2c631ee0	211	}
94fab3d1	212	print " ];\n" if $opt_json;
ee2793c1	213	print "}\n" if $opt_dot;
94fab3d1	214
b2fc39a5	215	$dbh->commit if $dbh;
b2fc39a5	216
2c631ee0	217	use Data::Dumper;
	218	warn Dumper(\$x);
	219	warn Dumper(\%seqn2node);
	220
	221	=for
	222	SV(PVAV) fill=1/1 [#1 @0]
	223	: +64 sv =64
	224	: +16 av_max =80
	225	: AVelem-> [#2 @1]
	226	: : SV(RV) [#3 @2]
	227	: : : +24 sv =104
	228	: : : RV-> [#4 @3]
	229	: : : : SV(PVAV) fill=-1/-1 [#5 @4]
	230	: : : : : +64 sv =168
	231	: AVelem-> [#6 @1]
	232	: : SV(IV) [#7 @2]
	233	: : : +24 sv =192
	234	192 at -e line 1.
	235	=cut
	236	__DATA__
	237	N 1 0 SV(PVAV) fill=1/1
	238	L 1 64 sv
	239	L 1 16 av_max
	240	N 2 1 AVelem->
	241	N 3 2 SV(RV)
	242	L 3 24 sv
	243	N 4 3 RV->
	244	N 5 4 SV(PVAV) fill=-1/-1
	245	L 5 64 sv
	246	N 6 1 AVelem->
	247	N 7 2 SV(IV)
	248	L 7 24 sv