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; |
b2fc39a5 |
10 | |
fc6614ee |
11 | use Getopt::Long; |
12 | |
13 | GetOptions( |
5aa3ad8e |
14 | 'text!' => \my $opt_text, |
15 | 'dot=s' => \my $opt_dot, |
b2fc39a5 |
16 | 'db=s' => \my $opt_db, |
e8f4c506 |
17 | 'verbose|v!' => \my $opt_verbose, |
18 | 'debug|d!' => \my $opt_debug, |
98128850 |
19 | 'showid!' => \my $opt_showid, |
fc6614ee |
20 | ) or exit 1; |
94fab3d1 |
21 | |
f60f09e5 |
22 | my $j = JSON::XS->new->ascii->pretty(0); |
23 | |
1915946b |
24 | my ($dbh, $node_ins_sth); |
25 | if ($opt_db) { |
26 | $dbh = DBI->connect("dbi:SQLite:dbname=$opt_db","","", { |
27 | RaiseError => 1, PrintError => 0, AutoCommit => 0 |
28 | }); |
29 | $dbh->do("PRAGMA synchronous = OFF"); |
30 | $dbh->do("DROP TABLE IF EXISTS node"); |
31 | $dbh->do(q{ |
32 | CREATE TABLE node ( |
33 | id integer primary key, |
34 | name text, |
35 | title text, |
98128850 |
36 | type integer, |
1915946b |
37 | depth integer, |
38 | parent_id integer, |
39 | |
40 | self_size integer, |
41 | kids_size integer, |
42 | kids_node_count integer, |
43 | child_ids text, |
44 | attr_json text, |
45 | leaves_json text |
46 | ) |
47 | }); |
48 | $node_ins_sth = $dbh->prepare(q{ |
98128850 |
49 | INSERT INTO node VALUES (?,?,?,?,?,?, ?,?,?,?,?,?) |
1915946b |
50 | }); |
51 | } |
b2fc39a5 |
52 | |
2c631ee0 |
53 | my @stack; |
54 | my %seqn2node; |
55 | |
7020702a |
56 | use HTML::Entities qw(encode_entities);; |
57 | my $dotnode = sub { |
58 | my $name = encode_entities(shift); |
59 | $name =~ s/"/\\"/g; |
7020702a |
60 | return '"'.$name.'"'; |
61 | }; |
ee2793c1 |
62 | |
ee2793c1 |
63 | |
5aa3ad8e |
64 | my $dot_fh; |
ee2793c1 |
65 | if ($opt_dot) { |
5aa3ad8e |
66 | open $dot_fh, ">$opt_dot"; |
67 | print $dot_fh "digraph {\n"; # } |
68 | print $dot_fh "graph [overlap=false]\n"; # target="???", URL="???" |
ee2793c1 |
69 | } |
70 | |
0741448c |
71 | sub fmt_size { |
72 | my $size = shift; |
73 | my $kb = $size / 1024; |
74 | return $size if $kb < 5; |
75 | return sprintf "%.1fKb", $kb if $kb < 1000; |
76 | return sprintf "%.1fMb", $kb/1024; |
77 | } |
78 | |
ee2793c1 |
79 | |
94fab3d1 |
80 | sub enter_node { |
81 | my $x = shift; |
ee2793c1 |
82 | if ($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}}; |
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 |
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}}, $x->{id}; |
2c631ee0 |
104 | } |
105 | # output |
106 | # ... |
ee2793c1 |
107 | if ($opt_dot) { |
1915946b |
108 | printf "// n%d parent=%s(type=%s)\n", $x->{id}, |
0741448c |
109 | $parent ? $parent->{id} : "", |
110 | $parent ? $parent->{type} : "" |
111 | if 0; |
1915946b |
112 | if ($x->{type} != 2) { |
0741448c |
113 | my $name = $x->{title} ? "\"$x->{title}\" $x->{name}" : $x->{name}; |
114 | |
115 | if ($x->{kids_size}) { |
116 | $name .= sprintf " %s+%s=%s", fmt_size($x->{self_size}), fmt_size($x->{kids_size}), fmt_size($x->{self_size}+$x->{kids_size}); |
117 | } |
118 | else { |
119 | $name .= sprintf " +%s", fmt_size($x->{self_size}); |
120 | } |
98128850 |
121 | $name .= " $x->{id}" if $opt_showid; |
0741448c |
122 | |
123 | my @node_attr = ( |
124 | sprintf("label=%s", $dotnode->($name)), |
125 | "id=$x->{id}", |
126 | ); |
127 | my @link_attr; |
128 | #if ($x->{name} eq 'hek') { push @node_attr, "shape=point"; push @node_attr, "labelfontsize=6"; } |
1915946b |
129 | if ($parent) { # probably a link |
1915946b |
130 | my $parent_id = $parent->{id}; |
0741448c |
131 | my @link_attr = ("id=$parent_id"); |
1915946b |
132 | if ($parent->{type} == 2) { # link |
133 | (my $link_name = $parent->{name}) =~ s/->$//; |
134 | push @link_attr, (sprintf "label=%s", $dotnode->($link_name)); |
135 | $parent_id = ($stack[-2]||die "panic")->{id}; |
136 | } |
5aa3ad8e |
137 | printf $dot_fh qq{n%d -> n%d [%s];\n}, |
1915946b |
138 | $parent_id, $x->{id}, join(",", @link_attr); |
139 | } |
5aa3ad8e |
140 | printf $dot_fh qq{n%d [ %s ];\n}, $x->{id}, join(",", @node_attr); |
1915946b |
141 | } |
142 | |
ee2793c1 |
143 | } |
b2fc39a5 |
144 | if ($dbh) { |
f60f09e5 |
145 | my $attr_json = $j->encode($x->{attr}); |
e78b28ca |
146 | my $leaves_json = $j->encode($x->{leaves}); |
b2fc39a5 |
147 | $node_ins_sth->execute( |
98128850 |
148 | $x->{id}, $x->{name}, $x->{title}, $x->{type}, $x->{depth}, $x->{parent_id}, |
b2fc39a5 |
149 | $x->{self_size}, $x->{kids_size}, $x->{kids_node_count}, |
f60f09e5 |
150 | $x->{child_id} ? join(",", @{$x->{child_id}}) : undef, |
e78b28ca |
151 | $attr_json, $leaves_json, |
b2fc39a5 |
152 | ); |
153 | # XXX attribs |
154 | } |
94fab3d1 |
155 | return; |
2c631ee0 |
156 | } |
157 | |
5aa3ad8e |
158 | my $indent = ": "; |
94fab3d1 |
159 | |
2c631ee0 |
160 | while (<>) { |
161 | chomp; |
b2fc39a5 |
162 | my ($type, $id, $val, $name, $extra) = split / /, $_, 5; |
ee2793c1 |
163 | if ($type =~ s/^-//) { # Node type ($val is depth) |
5aa3ad8e |
164 | printf "%s%s %s [#%d @%d]\n", $indent x $val, $name, $extra||'', $id, $val |
165 | if $opt_text; |
2c631ee0 |
166 | while ($val < @stack) { |
94fab3d1 |
167 | leave_node(my $x = pop @stack); |
e8f4c506 |
168 | warn "N $id d$val ends $x->{id} d$x->{depth}: size $x->{self_size}+$x->{kids_size}\n" |
169 | if $opt_verbose; |
2c631ee0 |
170 | } |
171 | die 1 if $stack[$val]; |
ee2793c1 |
172 | my $node = $stack[$val] = { id => $id, type => $type, name => $name, extra => $extra, attr => {}, leaves => {}, depth => $val, self_size=>0, kids_size=>0 }; |
94fab3d1 |
173 | enter_node($node); |
b2fc39a5 |
174 | $seqn2node{$id} = $node; |
2c631ee0 |
175 | } |
176 | elsif ($type eq "L") { # Leaf name and memory size |
b2fc39a5 |
177 | my $node = $seqn2node{$id} || die; |
2c631ee0 |
178 | $node->{leaves}{$name} += $val; |
5aa3ad8e |
179 | printf "%s+%d %s\n", $indent x ($node->{depth}+1), $val, $name |
180 | if $opt_text; |
2c631ee0 |
181 | } |
e8f4c506 |
182 | elsif (looks_like_number($type)) { # Attribute type, name and value |
b2fc39a5 |
183 | my $node = $seqn2node{$id} || die; |
e8f4c506 |
184 | my $attr = $node->{attr} || die; |
5aa3ad8e |
185 | printf "%s~%s %d [t%d]\n", $indent x ($node->{depth}+1), $name, $val, $type |
186 | if $opt_text; |
187 | if ($type == 1 or $type == 5) { # NPattr_NAME |
e8f4c506 |
188 | warn "Node $id already has attribute $type:$name (value $attr->{$type}{$name})\n" |
189 | if exists $attr->{$type}{$name}; |
190 | $attr->{$type}{$name} = $val || $id; |
e8f4c506 |
191 | $node->{title} = $name if $type == 1 and !$val; |
192 | } |
193 | elsif (2 <= $type and $type <= 4) { # NPattr_PAD* |
194 | warn "Node $id already has attribute $type:$name (value $attr->{$type}[$val])\n" |
195 | if defined $attr->{$type}[$val]; |
196 | $attr->{$type}[$val] = $name; |
197 | } |
198 | else { |
199 | warn "Invalid attribute type '$type' on line $. ($_)"; |
200 | } |
2c631ee0 |
201 | } |
202 | else { |
203 | warn "Invalid type '$type' on line $. ($_)"; |
e8f4c506 |
204 | next; |
2c631ee0 |
205 | } |
b2fc39a5 |
206 | $dbh->commit if $dbh and $id % 10_000 == 0; |
2c631ee0 |
207 | } |
208 | |
209 | my $x; |
210 | while (@stack > 1) { |
94fab3d1 |
211 | leave_node($x = pop @stack) while @stack; |
5aa3ad8e |
212 | warn "EOF ends $x->{id} d$x->{depth}: size $x->{self_size}+$x->{kids_size}\n" |
213 | if $opt_verbose; |
214 | } |
215 | |
216 | if ($dot_fh) { |
217 | print $dot_fh "}\n"; |
218 | close $dot_fh; |
219 | system("open -a Graphviz $opt_dot"); |
2c631ee0 |
220 | } |
94fab3d1 |
221 | |
b2fc39a5 |
222 | $dbh->commit if $dbh; |
223 | |
2c631ee0 |
224 | use Data::Dumper; |
5aa3ad8e |
225 | warn Dumper(\$x) if $opt_verbose; |
226 | warn Dumper(\%seqn2node) if %seqn2node; # should be empty |
2c631ee0 |
227 | |
228 | =for |
229 | SV(PVAV) fill=1/1 [#1 @0] |
230 | : +64 sv =64 |
231 | : +16 av_max =80 |
232 | : AVelem-> [#2 @1] |
233 | : : SV(RV) [#3 @2] |
234 | : : : +24 sv =104 |
235 | : : : RV-> [#4 @3] |
236 | : : : : SV(PVAV) fill=-1/-1 [#5 @4] |
237 | : : : : : +64 sv =168 |
238 | : AVelem-> [#6 @1] |
239 | : : SV(IV) [#7 @2] |
240 | : : : +24 sv =192 |
241 | 192 at -e line 1. |
242 | =cut |
243 | __DATA__ |
244 | N 1 0 SV(PVAV) fill=1/1 |
245 | L 1 64 sv |
246 | L 1 16 av_max |
247 | N 2 1 AVelem-> |
248 | N 3 2 SV(RV) |
249 | L 3 24 sv |
250 | N 4 3 RV-> |
251 | N 5 4 SV(PVAV) fill=-1/-1 |
252 | L 5 64 sv |
253 | N 6 1 AVelem-> |
254 | N 7 2 SV(IV) |
255 | L 7 24 sv |