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. |
d81c4128 |
6 | # |
d3b8a135 |
7 | # Needs to be generalized to support pluggable output formats. |
d81c4128 |
8 | # Actually it needs to be split so sizeme_store.pl only does the store |
9 | # and another program drives the output with plugins. |
d3b8a135 |
10 | # Making nodes into (lightweight fast) objects would be smart. |
11 | # Tests would be even smarter! |
12 | # |
13 | # When working on this code it's important to have a sense of the flow. |
14 | # Specifically the way that depth drives the completion of nodes. |
15 | # It's a depth-first stream processing machine, which only ever holds |
16 | # a single stack of the currently incomplete nodes, which is always the same as |
17 | # the current depth. I.e., when a node of depth N arrives, all nodes >N are |
18 | # popped off the stack and 'completed', each rippling data up to its parent. |
19 | |
2c631ee0 |
20 | use strict; |
21 | use warnings; |
5aa3ad8e |
22 | use autodie; |
2c631ee0 |
23 | |
e8f4c506 |
24 | use DBI qw(looks_like_number); |
b2fc39a5 |
25 | use DBD::SQLite; |
f60f09e5 |
26 | use JSON::XS; |
de73b186 |
27 | use Devel::Dwarn; |
d3b8a135 |
28 | use HTML::Entities qw(encode_entities);; |
5e2e22f3 |
29 | use Data::Dumper; |
fc6614ee |
30 | use Getopt::Long; |
5e2e22f3 |
31 | use Carp qw(carp croak confess); |
fc6614ee |
32 | |
de73b186 |
33 | # XXX import these from the XS code |
34 | use constant NPtype_NAME => 0x01; |
35 | use constant NPtype_LINK => 0x02; |
36 | use constant NPtype_SV => 0x03; |
37 | use constant NPtype_MAGIC => 0x04; |
38 | use constant NPtype_OP => 0x05; |
39 | |
40 | use constant NPattr_LEAFSIZE => 0x00; |
41 | use constant NPattr_NAME => 0x01; |
42 | use constant NPattr_PADFAKE => 0x02; |
43 | use constant NPattr_PADNAME => 0x03; |
44 | use constant NPattr_PADTMP => 0x04; |
45 | use constant NPattr_NOTE => 0x05; |
68cafb30 |
46 | use constant NPattr_PRE_ATTR => 0x06; |
09c6d3bb |
47 | my @attr_type_name = (qw(size NAME PADFAKE my PADTMP NOTE PREATTR)); # XXX get from XS in some way |
de73b186 |
48 | |
49 | |
fc6614ee |
50 | GetOptions( |
5aa3ad8e |
51 | 'text!' => \my $opt_text, |
52 | 'dot=s' => \my $opt_dot, |
b2fc39a5 |
53 | 'db=s' => \my $opt_db, |
e8f4c506 |
54 | 'verbose|v!' => \my $opt_verbose, |
55 | 'debug|d!' => \my $opt_debug, |
98128850 |
56 | 'showid!' => \my $opt_showid, |
fc6614ee |
57 | ) or exit 1; |
94fab3d1 |
58 | |
5e2e22f3 |
59 | $| = 1 if $opt_debug; |
0e977dbc |
60 | my $run_size = 0; |
61 | my $total_size = 0; |
62 | |
f60f09e5 |
63 | my $j = JSON::XS->new->ascii->pretty(0); |
64 | |
1915946b |
65 | my ($dbh, $node_ins_sth); |
66 | if ($opt_db) { |
67 | $dbh = DBI->connect("dbi:SQLite:dbname=$opt_db","","", { |
68 | RaiseError => 1, PrintError => 0, AutoCommit => 0 |
69 | }); |
70 | $dbh->do("PRAGMA synchronous = OFF"); |
1915946b |
71 | } |
b2fc39a5 |
72 | |
2c631ee0 |
73 | my @stack; |
74 | my %seqn2node; |
75 | |
7020702a |
76 | my $dotnode = sub { |
77 | my $name = encode_entities(shift); |
78 | $name =~ s/"/\\"/g; |
7020702a |
79 | return '"'.$name.'"'; |
80 | }; |
ee2793c1 |
81 | |
ee2793c1 |
82 | |
5aa3ad8e |
83 | my $dot_fh; |
ee2793c1 |
84 | |
0741448c |
85 | sub fmt_size { |
86 | my $size = shift; |
87 | my $kb = $size / 1024; |
88 | return $size if $kb < 5; |
89 | return sprintf "%.1fKb", $kb if $kb < 1000; |
90 | return sprintf "%.1fMb", $kb/1024; |
91 | } |
92 | |
ee2793c1 |
93 | |
94fab3d1 |
94 | sub enter_node { |
95 | my $x = shift; |
09c6d3bb |
96 | warn ">> enter_node $x->{id}\n" if $opt_debug; |
de73b186 |
97 | |
98 | my $parent = $stack[-1]; |
99 | if ($parent) { |
100 | |
37836f2a |
101 | if ($x->{name} eq 'AVelem' and $parent->{name} eq 'SV(PVAV)') { |
09c6d3bb |
102 | my $index = $x->{attr}{+NPattr_NOTE}{i}; |
eda23e24 |
103 | #Dwarn $x->{attr}; |
104 | #Dwarn $index; |
68cafb30 |
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') { |
68cafb30 |
107 | my $padnames = $cvpl->{_cached}{padnames} ||= do { |
108 | my @names = @{ $cvpl->{attr}{+NPattr_PADNAME} || []}; |
109 | $_ = "my(".($_||'').")" for @names; |
110 | $names[0] = '@_'; |
111 | \@names; |
112 | }; |
fd0de560 |
113 | $x->{name} = (defined $index and $padnames->[$index]) || "?"; |
68cafb30 |
114 | $x->{name} =~ s/my\(SVs_PADTMP\)/PADTMP/; # XXX hack for neatness |
115 | } |
116 | else { |
fd0de560 |
117 | $x->{name} = "[$index]" if defined $index; |
de73b186 |
118 | } |
119 | } |
09c6d3bb |
120 | |
ee2793c1 |
121 | } |
de73b186 |
122 | |
123 | return $x; |
94fab3d1 |
124 | } |
125 | |
de73b186 |
126 | |
94fab3d1 |
127 | sub leave_node { |
128 | my $x = shift; |
5e2e22f3 |
129 | confess unless defined $x->{id}; |
09c6d3bb |
130 | warn "<< leave_node $x->{id}\n" if $opt_debug; |
b2fc39a5 |
131 | delete $seqn2node{$x->{id}}; |
ee2793c1 |
132 | |
d3b8a135 |
133 | my $self_size = 0; $self_size += $_ for values %{$x->{leaves}}; |
94fab3d1 |
134 | $x->{self_size} = $self_size; |
ee2793c1 |
135 | |
09c6d3bb |
136 | if ($x->{name} eq 'AVelem') { |
137 | my $index = $x->{attr}{+NPattr_NOTE}{i}; |
138 | $x->{name} = "[$index]" if defined $index; |
139 | } |
140 | |
ee2793c1 |
141 | my $parent = $stack[-1]; |
142 | if ($parent) { |
2c631ee0 |
143 | # link to parent |
5a78486c |
144 | $x->{parent_id} = $parent->{id}; |
2c631ee0 |
145 | # accumulate into parent |
146 | $parent->{kids_node_count} += 1 + ($x->{kids_node_count}||0); |
94fab3d1 |
147 | $parent->{kids_size} += $self_size + $x->{kids_size}; |
5a78486c |
148 | push @{$parent->{child_id}}, $x->{id}; |
2c631ee0 |
149 | } |
d81c4128 |
150 | else { |
151 | $x->{kids_node_count} ||= 0; |
152 | } |
de73b186 |
153 | |
2c631ee0 |
154 | # output |
155 | # ... |
ee2793c1 |
156 | if ($opt_dot) { |
1915946b |
157 | printf "// n%d parent=%s(type=%s)\n", $x->{id}, |
0741448c |
158 | $parent ? $parent->{id} : "", |
159 | $parent ? $parent->{type} : "" |
160 | if 0; |
09c6d3bb |
161 | |
de73b186 |
162 | if ($x->{type} != NPtype_LINK) { |
0741448c |
163 | my $name = $x->{title} ? "\"$x->{title}\" $x->{name}" : $x->{name}; |
164 | |
165 | if ($x->{kids_size}) { |
166 | $name .= sprintf " %s+%s=%s", fmt_size($x->{self_size}), fmt_size($x->{kids_size}), fmt_size($x->{self_size}+$x->{kids_size}); |
167 | } |
168 | else { |
169 | $name .= sprintf " +%s", fmt_size($x->{self_size}); |
170 | } |
f91a90e9 |
171 | $name .= " #$x->{id}" if $opt_showid; |
0741448c |
172 | |
173 | my @node_attr = ( |
174 | sprintf("label=%s", $dotnode->($name)), |
175 | "id=$x->{id}", |
176 | ); |
09c6d3bb |
177 | printf $dot_fh qq{n%d [ %s ];\n}, $x->{id}, join(",", @node_attr); |
178 | } |
179 | else { # NPtype_LINK |
d81c4128 |
180 | my @kids = @{$x->{child_id}||[]}; |
09c6d3bb |
181 | die "panic: NPtype_LINK has more than one child: @kids" |
182 | if @kids > 1; |
183 | for my $child_id (@kids) { # wouldn't work right, eg id= attr |
184 | #die Dwarn $x; |
185 | my @link_attr = ("id=$x->{id}"); |
186 | (my $link_name = $x->{name}) =~ s/->$//; |
d81c4128 |
187 | $link_name .= " #$x->{id}" if $opt_showid; |
09c6d3bb |
188 | push @link_attr, (sprintf "label=%s", $dotnode->($link_name)); |
5aa3ad8e |
189 | printf $dot_fh qq{n%d -> n%d [%s];\n}, |
09c6d3bb |
190 | $x->{parent_id}, $child_id, join(",", @link_attr); |
1915946b |
191 | } |
1915946b |
192 | } |
193 | |
ee2793c1 |
194 | } |
b2fc39a5 |
195 | if ($dbh) { |
f60f09e5 |
196 | my $attr_json = $j->encode($x->{attr}); |
e78b28ca |
197 | my $leaves_json = $j->encode($x->{leaves}); |
b2fc39a5 |
198 | $node_ins_sth->execute( |
98128850 |
199 | $x->{id}, $x->{name}, $x->{title}, $x->{type}, $x->{depth}, $x->{parent_id}, |
b2fc39a5 |
200 | $x->{self_size}, $x->{kids_size}, $x->{kids_node_count}, |
f60f09e5 |
201 | $x->{child_id} ? join(",", @{$x->{child_id}}) : undef, |
e78b28ca |
202 | $attr_json, $leaves_json, |
b2fc39a5 |
203 | ); |
204 | # XXX attribs |
205 | } |
d3b8a135 |
206 | |
207 | return $x; |
2c631ee0 |
208 | } |
209 | |
5aa3ad8e |
210 | my $indent = ": "; |
94fab3d1 |
211 | |
2c631ee0 |
212 | while (<>) { |
d81c4128 |
213 | warn "\t\t\t\t== $_" if $opt_debug; |
2c631ee0 |
214 | chomp; |
de73b186 |
215 | |
b2fc39a5 |
216 | my ($type, $id, $val, $name, $extra) = split / /, $_, 5; |
de73b186 |
217 | |
ee2793c1 |
218 | if ($type =~ s/^-//) { # Node type ($val is depth) |
d3b8a135 |
219 | |
65b2cf7d |
220 | printf "%s%s%s %s [#%d @%d]\n", $indent x $val, $name, |
221 | ($type == NPtype_LINK) ? "->" : "", |
222 | $extra||'', $id, $val |
5aa3ad8e |
223 | if $opt_text; |
d3b8a135 |
224 | |
225 | # this is the core driving logic |
2c631ee0 |
226 | while ($val < @stack) { |
d3b8a135 |
227 | my $x = leave_node(pop @stack); |
e8f4c506 |
228 | warn "N $id d$val ends $x->{id} d$x->{depth}: size $x->{self_size}+$x->{kids_size}\n" |
229 | if $opt_verbose; |
2c631ee0 |
230 | } |
c5078bcb |
231 | die "panic: stack already has item at depth $val" |
232 | if $stack[$val]; |
5e2e22f3 |
233 | die "Depth out of sync\n" if $val != @stack; |
de73b186 |
234 | my $node = enter_node({ |
235 | id => $id, type => $type, name => $name, extra => $extra, |
957d176b |
236 | attr => { }, leaves => {}, depth => $val, self_size=>0, kids_size=>0 |
de73b186 |
237 | }); |
238 | $stack[$val] = $node; |
b2fc39a5 |
239 | $seqn2node{$id} = $node; |
2c631ee0 |
240 | } |
d3b8a135 |
241 | |
de73b186 |
242 | # --- Leaf name and memory size |
243 | elsif ($type eq "L") { |
b2fc39a5 |
244 | my $node = $seqn2node{$id} || die; |
2c631ee0 |
245 | $node->{leaves}{$name} += $val; |
0e977dbc |
246 | $run_size += $val; |
247 | printf "%s+%d=%d %s\n", $indent x ($node->{depth}+1), $val, $run_size, $name |
5aa3ad8e |
248 | if $opt_text; |
2c631ee0 |
249 | } |
d3b8a135 |
250 | |
251 | # --- Attribute type, name and value (all rather hackish) |
de73b186 |
252 | elsif (looks_like_number($type)) { |
b2fc39a5 |
253 | my $node = $seqn2node{$id} || die; |
e8f4c506 |
254 | my $attr = $node->{attr} || die; |
de73b186 |
255 | |
68cafb30 |
256 | # attributes where the string is a key (or always empty and the type is the key) |
957d176b |
257 | if ($type == NPattr_NAME or $type == NPattr_NOTE) { |
de73b186 |
258 | printf "%s~%s(%s) %d [t%d]\n", $indent x ($node->{depth}+1), $attr_type_name[$type], $name, $val, $type |
259 | if $opt_text; |
e8f4c506 |
260 | warn "Node $id already has attribute $type:$name (value $attr->{$type}{$name})\n" |
261 | if exists $attr->{$type}{$name}; |
09c6d3bb |
262 | $attr->{$type}{$name} = $val; |
eda23e24 |
263 | #Dwarn $attr; |
09c6d3bb |
264 | $node->{title} = $name if $type == NPattr_NAME and !$val; # XXX hack |
e8f4c506 |
265 | } |
68cafb30 |
266 | # attributes where the number is a key (or always zero) |
de73b186 |
267 | elsif (NPattr_PADFAKE==$type or NPattr_PADTMP==$type or NPattr_PADNAME==$type) { |
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}[$val])\n" |
271 | if defined $attr->{$type}[$val]; |
de73b186 |
272 | $attr->{+NPattr_PADNAME}[$val] = $name; # store all as NPattr_PADNAME |
e8f4c506 |
273 | } |
274 | else { |
de73b186 |
275 | printf "%s~%s %d [t%d]\n", $indent x ($node->{depth}+1), $name, $val, $type |
276 | if $opt_text; |
e8f4c506 |
277 | warn "Invalid attribute type '$type' on line $. ($_)"; |
278 | } |
2c631ee0 |
279 | } |
5e2e22f3 |
280 | elsif ($type eq 'S') { # start of a run |
281 | die "Unexpected start token" if @stack; |
282 | if ($opt_dot) { |
283 | open $dot_fh, ">$opt_dot"; |
284 | print $dot_fh "digraph {\n"; # } |
285 | print $dot_fh "graph [overlap=false]\n"; # target="???", URL="???" |
286 | } |
287 | if ($dbh) { |
288 | # XXX add a size_run table records each run |
289 | # XXX pick a table name to store the run nodes in |
290 | #$run_ins_sth->execute( |
291 | my $table = "node"; |
292 | $dbh->do("DROP TABLE IF EXISTS $table"); |
293 | $dbh->do(qq{ |
294 | CREATE TABLE $table ( |
295 | id integer primary key, |
296 | name text, |
297 | title text, |
298 | type integer, |
299 | depth integer, |
300 | parent_id integer, |
301 | |
302 | self_size integer, |
303 | kids_size integer, |
304 | kids_node_count integer, |
305 | child_ids text, |
306 | attr_json text, |
307 | leaves_json text |
308 | ) |
309 | }); |
310 | $node_ins_sth = $dbh->prepare(qq{ |
311 | INSERT INTO $table VALUES (?,?,?,?,?,?, ?,?,?,?,?,?) |
312 | }); |
313 | } |
314 | } |
315 | elsif ($type eq 'E') { # end of a run |
316 | |
317 | my $top = $stack[0]; # grab top node before we pop all the nodes |
318 | leave_node(pop @stack) while @stack; |
c99e1fe8 |
319 | |
eda23e24 |
320 | # if nothing output (ie size(undef)) |
321 | $top ||= { self_size=>0, kids_size=>0, kids_node_count=>0 }; |
322 | |
5e2e22f3 |
323 | my $top_size = $top->{self_size}+$top->{kids_size}; |
324 | |
eda23e24 |
325 | printf "Stored %d nodes totalling %s [lines=%d size=%d write=%.2fs]\n", |
326 | 1+$top->{kids_node_count}, fmt_size($top_size), |
327 | $., $top_size, $val; |
c99e1fe8 |
328 | # the duration here ($val) is from Devel::SizeMe perspective |
329 | # ie doesn't include time to read file/pipe and commit to database. |
5e2e22f3 |
330 | |
331 | if ($opt_verbose or $run_size != $top_size) { |
332 | warn "EOF ends $top->{id} d$top->{depth}: size $top->{self_size}+$top->{kids_size}\n"; |
333 | warn Dumper($top); |
334 | } |
335 | die "panic: seqn2node should be empty ". Dumper(\%seqn2node) |
336 | if %seqn2node; |
5e2e22f3 |
337 | |
338 | if ($dot_fh) { |
339 | print $dot_fh "}\n"; |
340 | close $dot_fh; |
341 | system("open -a Graphviz $opt_dot") if $^O eq 'darwin'; # OSX |
342 | } |
343 | |
344 | $dbh->commit if $dbh; |
345 | } |
2c631ee0 |
346 | else { |
347 | warn "Invalid type '$type' on line $. ($_)"; |
e8f4c506 |
348 | next; |
2c631ee0 |
349 | } |
d3b8a135 |
350 | |
b2fc39a5 |
351 | $dbh->commit if $dbh and $id % 10_000 == 0; |
2c631ee0 |
352 | } |
5e2e22f3 |
353 | die "EOF without end token" if @stack; |
b2fc39a5 |
354 | |
2c631ee0 |
355 | |
d3b8a135 |
356 | =for This is out of date but gives you an idea of the data and stream |
357 | |
2c631ee0 |
358 | SV(PVAV) fill=1/1 [#1 @0] |
359 | : +64 sv =64 |
360 | : +16 av_max =80 |
361 | : AVelem-> [#2 @1] |
362 | : : SV(RV) [#3 @2] |
363 | : : : +24 sv =104 |
364 | : : : RV-> [#4 @3] |
365 | : : : : SV(PVAV) fill=-1/-1 [#5 @4] |
366 | : : : : : +64 sv =168 |
367 | : AVelem-> [#6 @1] |
368 | : : SV(IV) [#7 @2] |
369 | : : : +24 sv =192 |
370 | 192 at -e line 1. |
371 | =cut |
372 | __DATA__ |
373 | N 1 0 SV(PVAV) fill=1/1 |
374 | L 1 64 sv |
375 | L 1 16 av_max |
376 | N 2 1 AVelem-> |
377 | N 3 2 SV(RV) |
378 | L 3 24 sv |
379 | N 4 3 RV-> |
380 | N 5 4 SV(PVAV) fill=-1/-1 |
381 | L 5 64 sv |
382 | N 6 1 AVelem-> |
383 | N 7 2 SV(IV) |
384 | L 7 24 sv |