Commit | Line | Data |
5aa3ad8e |
1 | #!/usr/bin/env perl |
2c631ee0 |
2 | |
d3b8a135 |
3 | # Read the raw memory data from Devel::Memory and process the tree |
4 | # (as a stack, propagating data such as totals, up the tree). |
5 | # Output completed nodes in the request formats. |
6 | # Needs to be generalized to support pluggable output formats. |
7 | # Making nodes into (lightweight fast) objects would be smart. |
8 | # Tests would be even smarter! |
9 | # |
10 | # When working on this code it's important to have a sense of the flow. |
11 | # Specifically the way that depth drives the completion of nodes. |
12 | # It's a depth-first stream processing machine, which only ever holds |
13 | # a single stack of the currently incomplete nodes, which is always the same as |
14 | # the current depth. I.e., when a node of depth N arrives, all nodes >N are |
15 | # popped off the stack and 'completed', each rippling data up to its parent. |
16 | |
2c631ee0 |
17 | use strict; |
18 | use warnings; |
5aa3ad8e |
19 | use autodie; |
2c631ee0 |
20 | |
e8f4c506 |
21 | use DBI qw(looks_like_number); |
b2fc39a5 |
22 | use DBD::SQLite; |
f60f09e5 |
23 | use JSON::XS; |
de73b186 |
24 | use Devel::Dwarn; |
d3b8a135 |
25 | use HTML::Entities qw(encode_entities);; |
b2fc39a5 |
26 | |
fc6614ee |
27 | use Getopt::Long; |
28 | |
de73b186 |
29 | # XXX import these from the XS code |
30 | use constant NPtype_NAME => 0x01; |
31 | use constant NPtype_LINK => 0x02; |
32 | use constant NPtype_SV => 0x03; |
33 | use constant NPtype_MAGIC => 0x04; |
34 | use constant NPtype_OP => 0x05; |
35 | |
36 | use constant NPattr_LEAFSIZE => 0x00; |
37 | use constant NPattr_NAME => 0x01; |
38 | use constant NPattr_PADFAKE => 0x02; |
39 | use constant NPattr_PADNAME => 0x03; |
40 | use constant NPattr_PADTMP => 0x04; |
41 | use constant NPattr_NOTE => 0x05; |
68cafb30 |
42 | use constant NPattr_PRE_ATTR => 0x06; |
de73b186 |
43 | |
44 | |
fc6614ee |
45 | GetOptions( |
5aa3ad8e |
46 | 'text!' => \my $opt_text, |
47 | 'dot=s' => \my $opt_dot, |
b2fc39a5 |
48 | 'db=s' => \my $opt_db, |
e8f4c506 |
49 | 'verbose|v!' => \my $opt_verbose, |
50 | 'debug|d!' => \my $opt_debug, |
98128850 |
51 | 'showid!' => \my $opt_showid, |
fc6614ee |
52 | ) or exit 1; |
94fab3d1 |
53 | |
f60f09e5 |
54 | my $j = JSON::XS->new->ascii->pretty(0); |
55 | |
1915946b |
56 | my ($dbh, $node_ins_sth); |
57 | if ($opt_db) { |
58 | $dbh = DBI->connect("dbi:SQLite:dbname=$opt_db","","", { |
59 | RaiseError => 1, PrintError => 0, AutoCommit => 0 |
60 | }); |
61 | $dbh->do("PRAGMA synchronous = OFF"); |
62 | $dbh->do("DROP TABLE IF EXISTS node"); |
63 | $dbh->do(q{ |
64 | CREATE TABLE node ( |
65 | id integer primary key, |
66 | name text, |
67 | title text, |
98128850 |
68 | type integer, |
1915946b |
69 | depth integer, |
70 | parent_id integer, |
71 | |
72 | self_size integer, |
73 | kids_size integer, |
74 | kids_node_count integer, |
75 | child_ids text, |
76 | attr_json text, |
77 | leaves_json text |
78 | ) |
79 | }); |
80 | $node_ins_sth = $dbh->prepare(q{ |
98128850 |
81 | INSERT INTO node VALUES (?,?,?,?,?,?, ?,?,?,?,?,?) |
1915946b |
82 | }); |
83 | } |
b2fc39a5 |
84 | |
2c631ee0 |
85 | my @stack; |
86 | my %seqn2node; |
87 | |
7020702a |
88 | my $dotnode = sub { |
89 | my $name = encode_entities(shift); |
90 | $name =~ s/"/\\"/g; |
7020702a |
91 | return '"'.$name.'"'; |
92 | }; |
ee2793c1 |
93 | |
ee2793c1 |
94 | |
5aa3ad8e |
95 | my $dot_fh; |
ee2793c1 |
96 | if ($opt_dot) { |
5aa3ad8e |
97 | open $dot_fh, ">$opt_dot"; |
98 | print $dot_fh "digraph {\n"; # } |
99 | print $dot_fh "graph [overlap=false]\n"; # target="???", URL="???" |
ee2793c1 |
100 | } |
101 | |
0741448c |
102 | sub fmt_size { |
103 | my $size = shift; |
104 | my $kb = $size / 1024; |
105 | return $size if $kb < 5; |
106 | return sprintf "%.1fKb", $kb if $kb < 1000; |
107 | return sprintf "%.1fMb", $kb/1024; |
108 | } |
109 | |
ee2793c1 |
110 | |
94fab3d1 |
111 | sub enter_node { |
112 | my $x = shift; |
de73b186 |
113 | |
114 | my $parent = $stack[-1]; |
115 | if ($parent) { |
116 | |
37836f2a |
117 | if ($x->{name} eq 'AVelem' and $parent->{name} eq 'SV(PVAV)') { |
68cafb30 |
118 | my $index = $x->{attr}{index}; |
119 | # If node is an AVelem of a CvPADLIST propagate pad name to AVelem |
120 | if (@stack >= 4 and (my $cvpl = $stack[-4])->{name} eq 'CvPADLIST') { |
68cafb30 |
121 | my $padnames = $cvpl->{_cached}{padnames} ||= do { |
122 | my @names = @{ $cvpl->{attr}{+NPattr_PADNAME} || []}; |
123 | $_ = "my(".($_||'').")" for @names; |
124 | $names[0] = '@_'; |
125 | \@names; |
126 | }; |
68cafb30 |
127 | $x->{name} = $padnames->[$index] || "?"; |
128 | $x->{name} =~ s/my\(SVs_PADTMP\)/PADTMP/; # XXX hack for neatness |
129 | } |
130 | else { |
131 | $x->{name} = "[$index]"; |
de73b186 |
132 | } |
133 | } |
ee2793c1 |
134 | } |
de73b186 |
135 | |
136 | return $x; |
94fab3d1 |
137 | } |
138 | |
de73b186 |
139 | |
94fab3d1 |
140 | sub leave_node { |
141 | my $x = shift; |
b2fc39a5 |
142 | delete $seqn2node{$x->{id}}; |
ee2793c1 |
143 | |
d3b8a135 |
144 | my $self_size = 0; $self_size += $_ for values %{$x->{leaves}}; |
94fab3d1 |
145 | $x->{self_size} = $self_size; |
ee2793c1 |
146 | |
147 | my $parent = $stack[-1]; |
148 | if ($parent) { |
2c631ee0 |
149 | # link to parent |
5a78486c |
150 | $x->{parent_id} = $parent->{id}; |
2c631ee0 |
151 | # accumulate into parent |
152 | $parent->{kids_node_count} += 1 + ($x->{kids_node_count}||0); |
94fab3d1 |
153 | $parent->{kids_size} += $self_size + $x->{kids_size}; |
5a78486c |
154 | push @{$parent->{child_id}}, $x->{id}; |
2c631ee0 |
155 | } |
de73b186 |
156 | |
2c631ee0 |
157 | # output |
158 | # ... |
ee2793c1 |
159 | if ($opt_dot) { |
1915946b |
160 | printf "// n%d parent=%s(type=%s)\n", $x->{id}, |
0741448c |
161 | $parent ? $parent->{id} : "", |
162 | $parent ? $parent->{type} : "" |
163 | if 0; |
de73b186 |
164 | if ($x->{type} != NPtype_LINK) { |
0741448c |
165 | my $name = $x->{title} ? "\"$x->{title}\" $x->{name}" : $x->{name}; |
166 | |
167 | if ($x->{kids_size}) { |
168 | $name .= sprintf " %s+%s=%s", fmt_size($x->{self_size}), fmt_size($x->{kids_size}), fmt_size($x->{self_size}+$x->{kids_size}); |
169 | } |
170 | else { |
171 | $name .= sprintf " +%s", fmt_size($x->{self_size}); |
172 | } |
98128850 |
173 | $name .= " $x->{id}" if $opt_showid; |
0741448c |
174 | |
175 | my @node_attr = ( |
176 | sprintf("label=%s", $dotnode->($name)), |
177 | "id=$x->{id}", |
178 | ); |
179 | my @link_attr; |
180 | #if ($x->{name} eq 'hek') { push @node_attr, "shape=point"; push @node_attr, "labelfontsize=6"; } |
1915946b |
181 | if ($parent) { # probably a link |
1915946b |
182 | my $parent_id = $parent->{id}; |
0741448c |
183 | my @link_attr = ("id=$parent_id"); |
de73b186 |
184 | if ($parent->{type} == NPtype_LINK) { # link |
1915946b |
185 | (my $link_name = $parent->{name}) =~ s/->$//; |
186 | push @link_attr, (sprintf "label=%s", $dotnode->($link_name)); |
187 | $parent_id = ($stack[-2]||die "panic")->{id}; |
188 | } |
5aa3ad8e |
189 | printf $dot_fh qq{n%d -> n%d [%s];\n}, |
1915946b |
190 | $parent_id, $x->{id}, join(",", @link_attr); |
191 | } |
5aa3ad8e |
192 | printf $dot_fh qq{n%d [ %s ];\n}, $x->{id}, join(",", @node_attr); |
1915946b |
193 | } |
194 | |
ee2793c1 |
195 | } |
b2fc39a5 |
196 | if ($dbh) { |
f60f09e5 |
197 | my $attr_json = $j->encode($x->{attr}); |
e78b28ca |
198 | my $leaves_json = $j->encode($x->{leaves}); |
b2fc39a5 |
199 | $node_ins_sth->execute( |
98128850 |
200 | $x->{id}, $x->{name}, $x->{title}, $x->{type}, $x->{depth}, $x->{parent_id}, |
b2fc39a5 |
201 | $x->{self_size}, $x->{kids_size}, $x->{kids_node_count}, |
f60f09e5 |
202 | $x->{child_id} ? join(",", @{$x->{child_id}}) : undef, |
e78b28ca |
203 | $attr_json, $leaves_json, |
b2fc39a5 |
204 | ); |
205 | # XXX attribs |
206 | } |
d3b8a135 |
207 | |
208 | return $x; |
2c631ee0 |
209 | } |
210 | |
5aa3ad8e |
211 | my $indent = ": "; |
d3b8a135 |
212 | my @attr_type_name = (qw(size NAME PADFAKE my PADTMP NOTE)); # XXX get from XS in some way |
68cafb30 |
213 | my $pending_pre_attr = {}; |
94fab3d1 |
214 | |
2c631ee0 |
215 | while (<>) { |
216 | chomp; |
de73b186 |
217 | |
b2fc39a5 |
218 | my ($type, $id, $val, $name, $extra) = split / /, $_, 5; |
de73b186 |
219 | |
ee2793c1 |
220 | if ($type =~ s/^-//) { # Node type ($val is depth) |
d3b8a135 |
221 | |
65b2cf7d |
222 | printf "%s%s%s %s [#%d @%d]\n", $indent x $val, $name, |
223 | ($type == NPtype_LINK) ? "->" : "", |
224 | $extra||'', $id, $val |
5aa3ad8e |
225 | if $opt_text; |
d3b8a135 |
226 | |
227 | # this is the core driving logic |
2c631ee0 |
228 | while ($val < @stack) { |
d3b8a135 |
229 | my $x = leave_node(pop @stack); |
e8f4c506 |
230 | warn "N $id d$val ends $x->{id} d$x->{depth}: size $x->{self_size}+$x->{kids_size}\n" |
231 | if $opt_verbose; |
2c631ee0 |
232 | } |
c5078bcb |
233 | die "panic: stack already has item at depth $val" |
234 | if $stack[$val]; |
de73b186 |
235 | my $node = enter_node({ |
236 | id => $id, type => $type, name => $name, extra => $extra, |
68cafb30 |
237 | attr => { %$pending_pre_attr }, |
238 | leaves => {}, depth => $val, self_size=>0, kids_size=>0 |
de73b186 |
239 | }); |
68cafb30 |
240 | %$pending_pre_attr = (); |
de73b186 |
241 | $stack[$val] = $node; |
b2fc39a5 |
242 | $seqn2node{$id} = $node; |
2c631ee0 |
243 | } |
d3b8a135 |
244 | |
de73b186 |
245 | # --- Leaf name and memory size |
246 | elsif ($type eq "L") { |
b2fc39a5 |
247 | my $node = $seqn2node{$id} || die; |
2c631ee0 |
248 | $node->{leaves}{$name} += $val; |
5aa3ad8e |
249 | printf "%s+%d %s\n", $indent x ($node->{depth}+1), $val, $name |
250 | if $opt_text; |
2c631ee0 |
251 | } |
d3b8a135 |
252 | |
253 | # --- Attribute type, name and value (all rather hackish) |
de73b186 |
254 | elsif (looks_like_number($type)) { |
b2fc39a5 |
255 | my $node = $seqn2node{$id} || die; |
e8f4c506 |
256 | my $attr = $node->{attr} || die; |
de73b186 |
257 | |
68cafb30 |
258 | # attributes to queue up and apply to the next node |
259 | if (NPattr_PRE_ATTR == $type) { |
260 | $pending_pre_attr->{$name} = $val; |
261 | } |
262 | # attributes where the string is a key (or always empty and the type is the key) |
263 | elsif ($type == NPattr_NAME or $type == NPattr_NOTE) { |
de73b186 |
264 | printf "%s~%s(%s) %d [t%d]\n", $indent x ($node->{depth}+1), $attr_type_name[$type], $name, $val, $type |
265 | if $opt_text; |
e8f4c506 |
266 | warn "Node $id already has attribute $type:$name (value $attr->{$type}{$name})\n" |
267 | if exists $attr->{$type}{$name}; |
268 | $attr->{$type}{$name} = $val || $id; |
e8f4c506 |
269 | $node->{title} = $name if $type == 1 and !$val; |
270 | } |
68cafb30 |
271 | # attributes where the number is a key (or always zero) |
de73b186 |
272 | elsif (NPattr_PADFAKE==$type or NPattr_PADTMP==$type or NPattr_PADNAME==$type) { |
273 | printf "%s~%s('%s') %d [t%d]\n", $indent x ($node->{depth}+1), $attr_type_name[$type], $name, $val, $type |
274 | if $opt_text; |
e8f4c506 |
275 | warn "Node $id already has attribute $type:$name (value $attr->{$type}[$val])\n" |
276 | if defined $attr->{$type}[$val]; |
de73b186 |
277 | $attr->{+NPattr_PADNAME}[$val] = $name; # store all as NPattr_PADNAME |
e8f4c506 |
278 | } |
279 | else { |
de73b186 |
280 | printf "%s~%s %d [t%d]\n", $indent x ($node->{depth}+1), $name, $val, $type |
281 | if $opt_text; |
e8f4c506 |
282 | warn "Invalid attribute type '$type' on line $. ($_)"; |
283 | } |
2c631ee0 |
284 | } |
285 | else { |
286 | warn "Invalid type '$type' on line $. ($_)"; |
e8f4c506 |
287 | next; |
2c631ee0 |
288 | } |
d3b8a135 |
289 | |
b2fc39a5 |
290 | $dbh->commit if $dbh and $id % 10_000 == 0; |
2c631ee0 |
291 | } |
c5078bcb |
292 | my $top = $stack[0]; # grab top node before we pop all the nodes |
293 | leave_node(pop @stack) while @stack; |
d3b8a135 |
294 | |
295 | if ($opt_verbose) { |
296 | warn "EOF ends $top->{id} d$top->{depth}: size $top->{self_size}+$top->{kids_size}\n"; |
297 | warn Dumper($top); |
298 | } |
5aa3ad8e |
299 | |
300 | if ($dot_fh) { |
301 | print $dot_fh "}\n"; |
302 | close $dot_fh; |
55e6dd93 |
303 | system("open -a Graphviz $opt_dot") if $^O eq 'darwin'; # OSX |
2c631ee0 |
304 | } |
94fab3d1 |
305 | |
b2fc39a5 |
306 | $dbh->commit if $dbh; |
307 | |
2c631ee0 |
308 | use Data::Dumper; |
5aa3ad8e |
309 | warn Dumper(\%seqn2node) if %seqn2node; # should be empty |
2c631ee0 |
310 | |
d3b8a135 |
311 | =for This is out of date but gives you an idea of the data and stream |
312 | |
2c631ee0 |
313 | SV(PVAV) fill=1/1 [#1 @0] |
314 | : +64 sv =64 |
315 | : +16 av_max =80 |
316 | : AVelem-> [#2 @1] |
317 | : : SV(RV) [#3 @2] |
318 | : : : +24 sv =104 |
319 | : : : RV-> [#4 @3] |
320 | : : : : SV(PVAV) fill=-1/-1 [#5 @4] |
321 | : : : : : +64 sv =168 |
322 | : AVelem-> [#6 @1] |
323 | : : SV(IV) [#7 @2] |
324 | : : : +24 sv =192 |
325 | 192 at -e line 1. |
326 | =cut |
327 | __DATA__ |
328 | N 1 0 SV(PVAV) fill=1/1 |
329 | L 1 64 sv |
330 | L 1 16 av_max |
331 | N 2 1 AVelem-> |
332 | N 3 2 SV(RV) |
333 | L 3 24 sv |
334 | N 4 3 RV-> |
335 | N 5 4 SV(PVAV) fill=-1/-1 |
336 | L 5 64 sv |
337 | N 6 1 AVelem-> |
338 | N 7 2 SV(IV) |
339 | L 7 24 sv |