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; |
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; |
68cafb30 |
27 | use constant NPattr_PRE_ATTR => 0x06; |
de73b186 |
28 | |
29 | |
fc6614ee |
30 | GetOptions( |
5aa3ad8e |
31 | 'text!' => \my $opt_text, |
32 | 'dot=s' => \my $opt_dot, |
b2fc39a5 |
33 | 'db=s' => \my $opt_db, |
e8f4c506 |
34 | 'verbose|v!' => \my $opt_verbose, |
35 | 'debug|d!' => \my $opt_debug, |
98128850 |
36 | 'showid!' => \my $opt_showid, |
fc6614ee |
37 | ) or exit 1; |
94fab3d1 |
38 | |
f60f09e5 |
39 | my $j = JSON::XS->new->ascii->pretty(0); |
40 | |
1915946b |
41 | my ($dbh, $node_ins_sth); |
42 | if ($opt_db) { |
43 | $dbh = DBI->connect("dbi:SQLite:dbname=$opt_db","","", { |
44 | RaiseError => 1, PrintError => 0, AutoCommit => 0 |
45 | }); |
46 | $dbh->do("PRAGMA synchronous = OFF"); |
47 | $dbh->do("DROP TABLE IF EXISTS node"); |
48 | $dbh->do(q{ |
49 | CREATE TABLE node ( |
50 | id integer primary key, |
51 | name text, |
52 | title text, |
98128850 |
53 | type integer, |
1915946b |
54 | depth integer, |
55 | parent_id integer, |
56 | |
57 | self_size integer, |
58 | kids_size integer, |
59 | kids_node_count integer, |
60 | child_ids text, |
61 | attr_json text, |
62 | leaves_json text |
63 | ) |
64 | }); |
65 | $node_ins_sth = $dbh->prepare(q{ |
98128850 |
66 | INSERT INTO node VALUES (?,?,?,?,?,?, ?,?,?,?,?,?) |
1915946b |
67 | }); |
68 | } |
b2fc39a5 |
69 | |
2c631ee0 |
70 | my @stack; |
71 | my %seqn2node; |
72 | |
7020702a |
73 | use HTML::Entities qw(encode_entities);; |
74 | my $dotnode = sub { |
75 | my $name = encode_entities(shift); |
76 | $name =~ s/"/\\"/g; |
7020702a |
77 | return '"'.$name.'"'; |
78 | }; |
ee2793c1 |
79 | |
ee2793c1 |
80 | |
5aa3ad8e |
81 | my $dot_fh; |
ee2793c1 |
82 | if ($opt_dot) { |
5aa3ad8e |
83 | open $dot_fh, ">$opt_dot"; |
84 | print $dot_fh "digraph {\n"; # } |
85 | print $dot_fh "graph [overlap=false]\n"; # target="???", URL="???" |
ee2793c1 |
86 | } |
87 | |
0741448c |
88 | sub fmt_size { |
89 | my $size = shift; |
90 | my $kb = $size / 1024; |
91 | return $size if $kb < 5; |
92 | return sprintf "%.1fKb", $kb if $kb < 1000; |
93 | return sprintf "%.1fMb", $kb/1024; |
94 | } |
95 | |
ee2793c1 |
96 | |
94fab3d1 |
97 | sub enter_node { |
98 | my $x = shift; |
de73b186 |
99 | |
100 | my $parent = $stack[-1]; |
101 | if ($parent) { |
102 | |
37836f2a |
103 | if ($x->{name} eq 'AVelem' and $parent->{name} eq 'SV(PVAV)') { |
68cafb30 |
104 | my $index = $x->{attr}{index}; |
105 | # If node is an AVelem of a CvPADLIST propagate pad name to AVelem |
106 | if (@stack >= 4 and (my $cvpl = $stack[-4])->{name} eq 'CvPADLIST') { |
107 | # cache the pad names so we can eat them in order |
108 | my $padnames = $cvpl->{_cached}{padnames} ||= do { |
109 | my @names = @{ $cvpl->{attr}{+NPattr_PADNAME} || []}; |
110 | $_ = "my(".($_||'').")" for @names; |
111 | $names[0] = '@_'; |
112 | \@names; |
113 | }; |
114 | #die Dwarn $x; |
115 | $x->{name} = $padnames->[$index] || "?"; |
116 | $x->{name} =~ s/my\(SVs_PADTMP\)/PADTMP/; # XXX hack for neatness |
117 | } |
118 | else { |
119 | $x->{name} = "[$index]"; |
de73b186 |
120 | } |
121 | } |
ee2793c1 |
122 | } |
de73b186 |
123 | |
124 | return $x; |
94fab3d1 |
125 | } |
126 | |
de73b186 |
127 | |
94fab3d1 |
128 | sub leave_node { |
129 | my $x = shift; |
b2fc39a5 |
130 | delete $seqn2node{$x->{id}}; |
ee2793c1 |
131 | |
94fab3d1 |
132 | my $self_size = 0; $self_size += $_ for values %{$x->{leaves}}; |
133 | $x->{self_size} = $self_size; |
ee2793c1 |
134 | |
135 | my $parent = $stack[-1]; |
136 | if ($parent) { |
2c631ee0 |
137 | # link to parent |
5a78486c |
138 | $x->{parent_id} = $parent->{id}; |
2c631ee0 |
139 | # accumulate into parent |
140 | $parent->{kids_node_count} += 1 + ($x->{kids_node_count}||0); |
94fab3d1 |
141 | $parent->{kids_size} += $self_size + $x->{kids_size}; |
5a78486c |
142 | push @{$parent->{child_id}}, $x->{id}; |
2c631ee0 |
143 | } |
de73b186 |
144 | |
2c631ee0 |
145 | # output |
146 | # ... |
ee2793c1 |
147 | if ($opt_dot) { |
1915946b |
148 | printf "// n%d parent=%s(type=%s)\n", $x->{id}, |
0741448c |
149 | $parent ? $parent->{id} : "", |
150 | $parent ? $parent->{type} : "" |
151 | if 0; |
de73b186 |
152 | if ($x->{type} != NPtype_LINK) { |
0741448c |
153 | my $name = $x->{title} ? "\"$x->{title}\" $x->{name}" : $x->{name}; |
154 | |
155 | if ($x->{kids_size}) { |
156 | $name .= sprintf " %s+%s=%s", fmt_size($x->{self_size}), fmt_size($x->{kids_size}), fmt_size($x->{self_size}+$x->{kids_size}); |
157 | } |
158 | else { |
159 | $name .= sprintf " +%s", fmt_size($x->{self_size}); |
160 | } |
98128850 |
161 | $name .= " $x->{id}" if $opt_showid; |
0741448c |
162 | |
163 | my @node_attr = ( |
164 | sprintf("label=%s", $dotnode->($name)), |
165 | "id=$x->{id}", |
166 | ); |
167 | my @link_attr; |
168 | #if ($x->{name} eq 'hek') { push @node_attr, "shape=point"; push @node_attr, "labelfontsize=6"; } |
1915946b |
169 | if ($parent) { # probably a link |
1915946b |
170 | my $parent_id = $parent->{id}; |
0741448c |
171 | my @link_attr = ("id=$parent_id"); |
de73b186 |
172 | if ($parent->{type} == NPtype_LINK) { # link |
1915946b |
173 | (my $link_name = $parent->{name}) =~ s/->$//; |
174 | push @link_attr, (sprintf "label=%s", $dotnode->($link_name)); |
175 | $parent_id = ($stack[-2]||die "panic")->{id}; |
176 | } |
5aa3ad8e |
177 | printf $dot_fh qq{n%d -> n%d [%s];\n}, |
1915946b |
178 | $parent_id, $x->{id}, join(",", @link_attr); |
179 | } |
5aa3ad8e |
180 | printf $dot_fh qq{n%d [ %s ];\n}, $x->{id}, join(",", @node_attr); |
1915946b |
181 | } |
182 | |
ee2793c1 |
183 | } |
b2fc39a5 |
184 | if ($dbh) { |
f60f09e5 |
185 | my $attr_json = $j->encode($x->{attr}); |
e78b28ca |
186 | my $leaves_json = $j->encode($x->{leaves}); |
b2fc39a5 |
187 | $node_ins_sth->execute( |
98128850 |
188 | $x->{id}, $x->{name}, $x->{title}, $x->{type}, $x->{depth}, $x->{parent_id}, |
b2fc39a5 |
189 | $x->{self_size}, $x->{kids_size}, $x->{kids_node_count}, |
f60f09e5 |
190 | $x->{child_id} ? join(",", @{$x->{child_id}}) : undef, |
e78b28ca |
191 | $attr_json, $leaves_json, |
b2fc39a5 |
192 | ); |
193 | # XXX attribs |
194 | } |
94fab3d1 |
195 | return; |
2c631ee0 |
196 | } |
197 | |
5aa3ad8e |
198 | my $indent = ": "; |
de73b186 |
199 | my @attr_type_name = (qw(size NAME PADFAKE my PADTMP NOTE)); |
68cafb30 |
200 | my $pending_pre_attr = {}; |
94fab3d1 |
201 | |
2c631ee0 |
202 | while (<>) { |
203 | chomp; |
de73b186 |
204 | |
b2fc39a5 |
205 | my ($type, $id, $val, $name, $extra) = split / /, $_, 5; |
de73b186 |
206 | |
ee2793c1 |
207 | if ($type =~ s/^-//) { # Node type ($val is depth) |
65b2cf7d |
208 | printf "%s%s%s %s [#%d @%d]\n", $indent x $val, $name, |
209 | ($type == NPtype_LINK) ? "->" : "", |
210 | $extra||'', $id, $val |
5aa3ad8e |
211 | if $opt_text; |
2c631ee0 |
212 | while ($val < @stack) { |
94fab3d1 |
213 | leave_node(my $x = pop @stack); |
e8f4c506 |
214 | warn "N $id d$val ends $x->{id} d$x->{depth}: size $x->{self_size}+$x->{kids_size}\n" |
215 | if $opt_verbose; |
2c631ee0 |
216 | } |
c5078bcb |
217 | die "panic: stack already has item at depth $val" |
218 | if $stack[$val]; |
de73b186 |
219 | my $node = enter_node({ |
220 | id => $id, type => $type, name => $name, extra => $extra, |
68cafb30 |
221 | attr => { %$pending_pre_attr }, |
222 | leaves => {}, depth => $val, self_size=>0, kids_size=>0 |
de73b186 |
223 | }); |
68cafb30 |
224 | %$pending_pre_attr = (); |
de73b186 |
225 | $stack[$val] = $node; |
b2fc39a5 |
226 | $seqn2node{$id} = $node; |
2c631ee0 |
227 | } |
de73b186 |
228 | # --- Leaf name and memory size |
229 | elsif ($type eq "L") { |
b2fc39a5 |
230 | my $node = $seqn2node{$id} || die; |
2c631ee0 |
231 | $node->{leaves}{$name} += $val; |
5aa3ad8e |
232 | printf "%s+%d %s\n", $indent x ($node->{depth}+1), $val, $name |
233 | if $opt_text; |
2c631ee0 |
234 | } |
de73b186 |
235 | # --- Attribute type, name and value |
236 | elsif (looks_like_number($type)) { |
b2fc39a5 |
237 | my $node = $seqn2node{$id} || die; |
e8f4c506 |
238 | my $attr = $node->{attr} || die; |
de73b186 |
239 | |
68cafb30 |
240 | # attributes to queue up and apply to the next node |
241 | if (NPattr_PRE_ATTR == $type) { |
242 | $pending_pre_attr->{$name} = $val; |
243 | } |
244 | # attributes where the string is a key (or always empty and the type is the key) |
245 | elsif ($type == NPattr_NAME or $type == NPattr_NOTE) { |
de73b186 |
246 | printf "%s~%s(%s) %d [t%d]\n", $indent x ($node->{depth}+1), $attr_type_name[$type], $name, $val, $type |
247 | if $opt_text; |
e8f4c506 |
248 | warn "Node $id already has attribute $type:$name (value $attr->{$type}{$name})\n" |
249 | if exists $attr->{$type}{$name}; |
250 | $attr->{$type}{$name} = $val || $id; |
e8f4c506 |
251 | $node->{title} = $name if $type == 1 and !$val; |
252 | } |
68cafb30 |
253 | # attributes where the number is a key (or always zero) |
de73b186 |
254 | elsif (NPattr_PADFAKE==$type or NPattr_PADTMP==$type or NPattr_PADNAME==$type) { |
255 | printf "%s~%s('%s') %d [t%d]\n", $indent x ($node->{depth}+1), $attr_type_name[$type], $name, $val, $type |
256 | if $opt_text; |
e8f4c506 |
257 | warn "Node $id already has attribute $type:$name (value $attr->{$type}[$val])\n" |
258 | if defined $attr->{$type}[$val]; |
de73b186 |
259 | $attr->{+NPattr_PADNAME}[$val] = $name; # store all as NPattr_PADNAME |
e8f4c506 |
260 | } |
261 | else { |
de73b186 |
262 | printf "%s~%s %d [t%d]\n", $indent x ($node->{depth}+1), $name, $val, $type |
263 | if $opt_text; |
e8f4c506 |
264 | warn "Invalid attribute type '$type' on line $. ($_)"; |
265 | } |
2c631ee0 |
266 | } |
267 | else { |
268 | warn "Invalid type '$type' on line $. ($_)"; |
e8f4c506 |
269 | next; |
2c631ee0 |
270 | } |
b2fc39a5 |
271 | $dbh->commit if $dbh and $id % 10_000 == 0; |
2c631ee0 |
272 | } |
273 | |
c5078bcb |
274 | my $top = $stack[0]; # grab top node before we pop all the nodes |
275 | leave_node(pop @stack) while @stack; |
276 | warn "EOF ends $top->{id} d$top->{depth}: size $top->{self_size}+$top->{kids_size}\n" |
277 | if $opt_verbose; |
278 | warn Dumper($top) if $opt_verbose; |
5aa3ad8e |
279 | |
280 | if ($dot_fh) { |
281 | print $dot_fh "}\n"; |
282 | close $dot_fh; |
283 | system("open -a Graphviz $opt_dot"); |
2c631ee0 |
284 | } |
94fab3d1 |
285 | |
b2fc39a5 |
286 | $dbh->commit if $dbh; |
287 | |
2c631ee0 |
288 | use Data::Dumper; |
5aa3ad8e |
289 | warn Dumper(\%seqn2node) if %seqn2node; # should be empty |
2c631ee0 |
290 | |
291 | =for |
292 | SV(PVAV) fill=1/1 [#1 @0] |
293 | : +64 sv =64 |
294 | : +16 av_max =80 |
295 | : AVelem-> [#2 @1] |
296 | : : SV(RV) [#3 @2] |
297 | : : : +24 sv =104 |
298 | : : : RV-> [#4 @3] |
299 | : : : : SV(PVAV) fill=-1/-1 [#5 @4] |
300 | : : : : : +64 sv =168 |
301 | : AVelem-> [#6 @1] |
302 | : : SV(IV) [#7 @2] |
303 | : : : +24 sv =192 |
304 | 192 at -e line 1. |
305 | =cut |
306 | __DATA__ |
307 | N 1 0 SV(PVAV) fill=1/1 |
308 | L 1 64 sv |
309 | L 1 16 av_max |
310 | N 2 1 AVelem-> |
311 | N 3 2 SV(RV) |
312 | L 3 24 sv |
313 | N 4 3 RV-> |
314 | N 5 4 SV(PVAV) fill=-1/-1 |
315 | L 5 64 sv |
316 | N 6 1 AVelem-> |
317 | N 7 2 SV(IV) |
318 | L 7 24 sv |