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 | |
0e977dbc |
54 | my $run_size = 0; |
55 | my $total_size = 0; |
56 | |
f60f09e5 |
57 | my $j = JSON::XS->new->ascii->pretty(0); |
58 | |
1915946b |
59 | my ($dbh, $node_ins_sth); |
60 | if ($opt_db) { |
61 | $dbh = DBI->connect("dbi:SQLite:dbname=$opt_db","","", { |
62 | RaiseError => 1, PrintError => 0, AutoCommit => 0 |
63 | }); |
64 | $dbh->do("PRAGMA synchronous = OFF"); |
65 | $dbh->do("DROP TABLE IF EXISTS node"); |
66 | $dbh->do(q{ |
67 | CREATE TABLE node ( |
68 | id integer primary key, |
69 | name text, |
70 | title text, |
98128850 |
71 | type integer, |
1915946b |
72 | depth integer, |
73 | parent_id integer, |
74 | |
75 | self_size integer, |
76 | kids_size integer, |
77 | kids_node_count integer, |
78 | child_ids text, |
79 | attr_json text, |
80 | leaves_json text |
81 | ) |
82 | }); |
83 | $node_ins_sth = $dbh->prepare(q{ |
98128850 |
84 | INSERT INTO node VALUES (?,?,?,?,?,?, ?,?,?,?,?,?) |
1915946b |
85 | }); |
86 | } |
b2fc39a5 |
87 | |
2c631ee0 |
88 | my @stack; |
89 | my %seqn2node; |
90 | |
7020702a |
91 | my $dotnode = sub { |
92 | my $name = encode_entities(shift); |
93 | $name =~ s/"/\\"/g; |
7020702a |
94 | return '"'.$name.'"'; |
95 | }; |
ee2793c1 |
96 | |
ee2793c1 |
97 | |
5aa3ad8e |
98 | my $dot_fh; |
ee2793c1 |
99 | if ($opt_dot) { |
5aa3ad8e |
100 | open $dot_fh, ">$opt_dot"; |
101 | print $dot_fh "digraph {\n"; # } |
102 | print $dot_fh "graph [overlap=false]\n"; # target="???", URL="???" |
ee2793c1 |
103 | } |
104 | |
0741448c |
105 | sub fmt_size { |
106 | my $size = shift; |
107 | my $kb = $size / 1024; |
108 | return $size if $kb < 5; |
109 | return sprintf "%.1fKb", $kb if $kb < 1000; |
110 | return sprintf "%.1fMb", $kb/1024; |
111 | } |
112 | |
ee2793c1 |
113 | |
94fab3d1 |
114 | sub enter_node { |
115 | my $x = shift; |
de73b186 |
116 | |
117 | my $parent = $stack[-1]; |
118 | if ($parent) { |
119 | |
37836f2a |
120 | if ($x->{name} eq 'AVelem' and $parent->{name} eq 'SV(PVAV)') { |
68cafb30 |
121 | my $index = $x->{attr}{index}; |
122 | # If node is an AVelem of a CvPADLIST propagate pad name to AVelem |
123 | if (@stack >= 4 and (my $cvpl = $stack[-4])->{name} eq 'CvPADLIST') { |
68cafb30 |
124 | my $padnames = $cvpl->{_cached}{padnames} ||= do { |
125 | my @names = @{ $cvpl->{attr}{+NPattr_PADNAME} || []}; |
126 | $_ = "my(".($_||'').")" for @names; |
127 | $names[0] = '@_'; |
128 | \@names; |
129 | }; |
68cafb30 |
130 | $x->{name} = $padnames->[$index] || "?"; |
131 | $x->{name} =~ s/my\(SVs_PADTMP\)/PADTMP/; # XXX hack for neatness |
132 | } |
133 | else { |
134 | $x->{name} = "[$index]"; |
de73b186 |
135 | } |
136 | } |
ee2793c1 |
137 | } |
de73b186 |
138 | |
139 | return $x; |
94fab3d1 |
140 | } |
141 | |
de73b186 |
142 | |
94fab3d1 |
143 | sub leave_node { |
144 | my $x = shift; |
b2fc39a5 |
145 | delete $seqn2node{$x->{id}}; |
ee2793c1 |
146 | |
d3b8a135 |
147 | my $self_size = 0; $self_size += $_ for values %{$x->{leaves}}; |
94fab3d1 |
148 | $x->{self_size} = $self_size; |
ee2793c1 |
149 | |
150 | my $parent = $stack[-1]; |
151 | if ($parent) { |
2c631ee0 |
152 | # link to parent |
5a78486c |
153 | $x->{parent_id} = $parent->{id}; |
2c631ee0 |
154 | # accumulate into parent |
155 | $parent->{kids_node_count} += 1 + ($x->{kids_node_count}||0); |
94fab3d1 |
156 | $parent->{kids_size} += $self_size + $x->{kids_size}; |
5a78486c |
157 | push @{$parent->{child_id}}, $x->{id}; |
2c631ee0 |
158 | } |
de73b186 |
159 | |
2c631ee0 |
160 | # output |
161 | # ... |
ee2793c1 |
162 | if ($opt_dot) { |
1915946b |
163 | printf "// n%d parent=%s(type=%s)\n", $x->{id}, |
0741448c |
164 | $parent ? $parent->{id} : "", |
165 | $parent ? $parent->{type} : "" |
166 | if 0; |
de73b186 |
167 | if ($x->{type} != NPtype_LINK) { |
0741448c |
168 | my $name = $x->{title} ? "\"$x->{title}\" $x->{name}" : $x->{name}; |
169 | |
170 | if ($x->{kids_size}) { |
171 | $name .= sprintf " %s+%s=%s", fmt_size($x->{self_size}), fmt_size($x->{kids_size}), fmt_size($x->{self_size}+$x->{kids_size}); |
172 | } |
173 | else { |
174 | $name .= sprintf " +%s", fmt_size($x->{self_size}); |
175 | } |
98128850 |
176 | $name .= " $x->{id}" if $opt_showid; |
0741448c |
177 | |
178 | my @node_attr = ( |
179 | sprintf("label=%s", $dotnode->($name)), |
180 | "id=$x->{id}", |
181 | ); |
182 | my @link_attr; |
183 | #if ($x->{name} eq 'hek') { push @node_attr, "shape=point"; push @node_attr, "labelfontsize=6"; } |
1915946b |
184 | if ($parent) { # probably a link |
1915946b |
185 | my $parent_id = $parent->{id}; |
0741448c |
186 | my @link_attr = ("id=$parent_id"); |
de73b186 |
187 | if ($parent->{type} == NPtype_LINK) { # link |
1915946b |
188 | (my $link_name = $parent->{name}) =~ s/->$//; |
189 | push @link_attr, (sprintf "label=%s", $dotnode->($link_name)); |
190 | $parent_id = ($stack[-2]||die "panic")->{id}; |
191 | } |
5aa3ad8e |
192 | printf $dot_fh qq{n%d -> n%d [%s];\n}, |
1915946b |
193 | $parent_id, $x->{id}, join(",", @link_attr); |
194 | } |
5aa3ad8e |
195 | printf $dot_fh qq{n%d [ %s ];\n}, $x->{id}, join(",", @node_attr); |
1915946b |
196 | } |
197 | |
ee2793c1 |
198 | } |
b2fc39a5 |
199 | if ($dbh) { |
f60f09e5 |
200 | my $attr_json = $j->encode($x->{attr}); |
e78b28ca |
201 | my $leaves_json = $j->encode($x->{leaves}); |
b2fc39a5 |
202 | $node_ins_sth->execute( |
98128850 |
203 | $x->{id}, $x->{name}, $x->{title}, $x->{type}, $x->{depth}, $x->{parent_id}, |
b2fc39a5 |
204 | $x->{self_size}, $x->{kids_size}, $x->{kids_node_count}, |
f60f09e5 |
205 | $x->{child_id} ? join(",", @{$x->{child_id}}) : undef, |
e78b28ca |
206 | $attr_json, $leaves_json, |
b2fc39a5 |
207 | ); |
208 | # XXX attribs |
209 | } |
d3b8a135 |
210 | |
211 | return $x; |
2c631ee0 |
212 | } |
213 | |
5aa3ad8e |
214 | my $indent = ": "; |
d3b8a135 |
215 | my @attr_type_name = (qw(size NAME PADFAKE my PADTMP NOTE)); # XXX get from XS in some way |
68cafb30 |
216 | my $pending_pre_attr = {}; |
94fab3d1 |
217 | |
2c631ee0 |
218 | while (<>) { |
219 | chomp; |
de73b186 |
220 | |
b2fc39a5 |
221 | my ($type, $id, $val, $name, $extra) = split / /, $_, 5; |
de73b186 |
222 | |
ee2793c1 |
223 | if ($type =~ s/^-//) { # Node type ($val is depth) |
d3b8a135 |
224 | |
65b2cf7d |
225 | printf "%s%s%s %s [#%d @%d]\n", $indent x $val, $name, |
226 | ($type == NPtype_LINK) ? "->" : "", |
227 | $extra||'', $id, $val |
5aa3ad8e |
228 | if $opt_text; |
d3b8a135 |
229 | |
230 | # this is the core driving logic |
2c631ee0 |
231 | while ($val < @stack) { |
d3b8a135 |
232 | my $x = leave_node(pop @stack); |
e8f4c506 |
233 | warn "N $id d$val ends $x->{id} d$x->{depth}: size $x->{self_size}+$x->{kids_size}\n" |
234 | if $opt_verbose; |
2c631ee0 |
235 | } |
c5078bcb |
236 | die "panic: stack already has item at depth $val" |
237 | if $stack[$val]; |
de73b186 |
238 | my $node = enter_node({ |
239 | id => $id, type => $type, name => $name, extra => $extra, |
68cafb30 |
240 | attr => { %$pending_pre_attr }, |
241 | leaves => {}, depth => $val, self_size=>0, kids_size=>0 |
de73b186 |
242 | }); |
68cafb30 |
243 | %$pending_pre_attr = (); |
de73b186 |
244 | $stack[$val] = $node; |
b2fc39a5 |
245 | $seqn2node{$id} = $node; |
2c631ee0 |
246 | } |
d3b8a135 |
247 | |
de73b186 |
248 | # --- Leaf name and memory size |
249 | elsif ($type eq "L") { |
b2fc39a5 |
250 | my $node = $seqn2node{$id} || die; |
2c631ee0 |
251 | $node->{leaves}{$name} += $val; |
0e977dbc |
252 | $run_size += $val; |
253 | printf "%s+%d=%d %s\n", $indent x ($node->{depth}+1), $val, $run_size, $name |
5aa3ad8e |
254 | if $opt_text; |
2c631ee0 |
255 | } |
d3b8a135 |
256 | |
257 | # --- Attribute type, name and value (all rather hackish) |
de73b186 |
258 | elsif (looks_like_number($type)) { |
b2fc39a5 |
259 | my $node = $seqn2node{$id} || die; |
e8f4c506 |
260 | my $attr = $node->{attr} || die; |
de73b186 |
261 | |
68cafb30 |
262 | # attributes to queue up and apply to the next node |
263 | if (NPattr_PRE_ATTR == $type) { |
264 | $pending_pre_attr->{$name} = $val; |
265 | } |
266 | # attributes where the string is a key (or always empty and the type is the key) |
267 | elsif ($type == NPattr_NAME or $type == NPattr_NOTE) { |
de73b186 |
268 | printf "%s~%s(%s) %d [t%d]\n", $indent x ($node->{depth}+1), $attr_type_name[$type], $name, $val, $type |
269 | if $opt_text; |
e8f4c506 |
270 | warn "Node $id already has attribute $type:$name (value $attr->{$type}{$name})\n" |
271 | if exists $attr->{$type}{$name}; |
272 | $attr->{$type}{$name} = $val || $id; |
e8f4c506 |
273 | $node->{title} = $name if $type == 1 and !$val; |
274 | } |
68cafb30 |
275 | # attributes where the number is a key (or always zero) |
de73b186 |
276 | elsif (NPattr_PADFAKE==$type or NPattr_PADTMP==$type or NPattr_PADNAME==$type) { |
277 | printf "%s~%s('%s') %d [t%d]\n", $indent x ($node->{depth}+1), $attr_type_name[$type], $name, $val, $type |
278 | if $opt_text; |
e8f4c506 |
279 | warn "Node $id already has attribute $type:$name (value $attr->{$type}[$val])\n" |
280 | if defined $attr->{$type}[$val]; |
de73b186 |
281 | $attr->{+NPattr_PADNAME}[$val] = $name; # store all as NPattr_PADNAME |
e8f4c506 |
282 | } |
283 | else { |
de73b186 |
284 | printf "%s~%s %d [t%d]\n", $indent x ($node->{depth}+1), $name, $val, $type |
285 | if $opt_text; |
e8f4c506 |
286 | warn "Invalid attribute type '$type' on line $. ($_)"; |
287 | } |
2c631ee0 |
288 | } |
289 | else { |
290 | warn "Invalid type '$type' on line $. ($_)"; |
e8f4c506 |
291 | next; |
2c631ee0 |
292 | } |
d3b8a135 |
293 | |
b2fc39a5 |
294 | $dbh->commit if $dbh and $id % 10_000 == 0; |
2c631ee0 |
295 | } |
c5078bcb |
296 | my $top = $stack[0]; # grab top node before we pop all the nodes |
297 | leave_node(pop @stack) while @stack; |
0e977dbc |
298 | my $top_size = $top->{self_size}+$top->{kids_size}; |
299 | |
300 | printf "Stored %d nodes recording %s (%d)\n", |
301 | $top->{kids_node_count}, fmt_size($top_size), $top_size; |
d3b8a135 |
302 | |
0e977dbc |
303 | if ($opt_verbose or $run_size != $top_size) { |
d3b8a135 |
304 | warn "EOF ends $top->{id} d$top->{depth}: size $top->{self_size}+$top->{kids_size}\n"; |
305 | warn Dumper($top); |
306 | } |
5aa3ad8e |
307 | |
308 | if ($dot_fh) { |
309 | print $dot_fh "}\n"; |
310 | close $dot_fh; |
55e6dd93 |
311 | system("open -a Graphviz $opt_dot") if $^O eq 'darwin'; # OSX |
2c631ee0 |
312 | } |
94fab3d1 |
313 | |
b2fc39a5 |
314 | $dbh->commit if $dbh; |
315 | |
2c631ee0 |
316 | use Data::Dumper; |
5aa3ad8e |
317 | warn Dumper(\%seqn2node) if %seqn2node; # should be empty |
2c631ee0 |
318 | |
d3b8a135 |
319 | =for This is out of date but gives you an idea of the data and stream |
320 | |
2c631ee0 |
321 | SV(PVAV) fill=1/1 [#1 @0] |
322 | : +64 sv =64 |
323 | : +16 av_max =80 |
324 | : AVelem-> [#2 @1] |
325 | : : SV(RV) [#3 @2] |
326 | : : : +24 sv =104 |
327 | : : : RV-> [#4 @3] |
328 | : : : : SV(PVAV) fill=-1/-1 [#5 @4] |
329 | : : : : : +64 sv =168 |
330 | : AVelem-> [#6 @1] |
331 | : : SV(IV) [#7 @2] |
332 | : : : +24 sv =192 |
333 | 192 at -e line 1. |
334 | =cut |
335 | __DATA__ |
336 | N 1 0 SV(PVAV) fill=1/1 |
337 | L 1 64 sv |
338 | L 1 16 av_max |
339 | N 2 1 AVelem-> |
340 | N 3 2 SV(RV) |
341 | L 3 24 sv |
342 | N 4 3 RV-> |
343 | N 5 4 SV(PVAV) fill=-1/-1 |
344 | L 5 64 sv |
345 | N 6 1 AVelem-> |
346 | N 7 2 SV(IV) |
347 | L 7 24 sv |