[bpt/coccinelle.git] / scripts / stat_directory_complete.pl

#!/usr/bin/perl

use strict;
use diagnostics;

#use Data::Dumper;
#use Date::Manip qw(ParseDate UnixDate); #sudo apt-get install libdate-manip-perl
#use Date::Calc qw(Delta_Days); #sudo apt-get install libdate-calc-perl

#------------------------------------------------------------------------------
# Helpers 
#------------------------------------------------------------------------------

my $debug = 0; 

sub pr2 { print STDERR "@_\n"; }
sub pr { print "@_\n"; }
sub mylog { print STDERR "@_\n" if $debug; }

sub plural { 
  my ($n) = @_;
  $n > 1 ? "s" : "";
}

#------------------------------------------------------------------------------
# Globals
#------------------------------------------------------------------------------

my $ok = 0;
my $so = 0; #spatch ok 
my $fa = 0; #failed 
my $gu = 0; #gave up

my $nbfiles = 0;

my $maxtime = 0;
#my $mintime = 0;
my $sumtime = 0;

my $sumlinefiles = 0;

my $errors = 0;

my $sumlineP = 0; #whole git
my $sumlineP2 = 0;
my $sumlinePchange = 0;

my $spfile = "";
my $ruleno = "??";
my $sizeSP = 0;

my $cedescr = "";

my $numauthors = 0;
my $duration = 0; # in days 

my @cfiles = ();

#------------------------------------------------------------------------------
# SP
#------------------------------------------------------------------------------
$spfile = `make sp_file`;
chomp $spfile;

if($spfile =~ /(rule|mega|bt)(\d+)\.cocci/) { $ruleno = "$2"; }

#------------------------------------------------------------------------------
# CE
#------------------------------------------------------------------------------

#$cedescr = `make ce_descr`;
#chomp $cedescr;
#print STDERR (Dumper($cedescr));
open TMP, "Makefile" or die "no Makefile file ?";
while(<TMP>) { 
  if(/^(CE)?DESCRIPTION=["'](.*)["']/) {  $cedescr = $2; }
}


#$cedescr =~ s/\\/\\\\/g;
#$cedescr =~ s/\f/\\f/g;
#$cedescr =~ s/\t/\\t/g;

#------------------------------------------------------------------------------
# List c files 
#------------------------------------------------------------------------------
my $files = `make source_files`;
chomp $files;
@cfiles = split /\s+/, $files;

$nbfiles = scalar(@cfiles);

#------------------------------------------------------------------------------
# Size files (lines)
#------------------------------------------------------------------------------
map { 
  my ($linefile) = `wc -l $_`; 
  chomp $linefile;
  die "wierd wc output" unless $linefile =~ /^(\d+) /;
  $sumlinefiles += $1;
  mylog "filesize $_ $1";
} @cfiles;


#------------------------------------------------------------------------------
# Size SP
#------------------------------------------------------------------------------
$sizeSP = 
  `cat $spfile | perl -p -e "s/\\/\\/.*//g;" | grep -v '^[ \t]*\$' | wc -l`;
chomp $sizeSP;


#------------------------------------------------------------------------------
# Bugs
#------------------------------------------------------------------------------
if(!(-e "README")) { pr2 "no README file ?"; }
else {
  open TMP, "README" or die "no README file ?";
  while(<TMP>) { 
    if (/\[bug\]/ || /status\]\s*bug/ ||  /status\]\s*BUG/ ) { 

      # can also look for [semibug] but it's often related to [corrected_c] kind of pbs
      #|| /status\]\s*semi-bug/ 

      #pr2 "OLD BUG FORMAT: $_"; 
      $errors++ 
    }
  }
}


#------------------------------------------------------------------------------
# Size P (total)
#------------------------------------------------------------------------------

if(-e "gitinfo") { 
  ($sumlineP) = `cat gitinfo |wc -l`;
  chomp $sumlineP;
} else {
  pr2 "no GIT INFO?";
}

#------------------------------------------------------------------------------
# Number of authors and duration
#------------------------------------------------------------------------------


if(-e "gitinfo") { 
  
  open TMP, "gitinfo" or die "no gitinfo file ?";

  #for authors
  my $h = {};

  #for duration
  my @mindate = ();
  my @maxdate = ();
  my $nodateyet = 1;

  while(<TMP>) { 
    #can also do: egrep "^Author" gitinfo | sort | uniq | wc -l
    if (/^Author: (.*)/) {
      $h->{$1}++;
    }

#    if(/^Date: (.*) ([-+]\d+)?/) {
#      my $date = ParseDate($1);
#      if (!$date) { die "bad date" } 
#      else {
#        my ($year, $month, $day) = UnixDate($date, "%Y", "%m", "%d");
#        my @current = ($year, $month, $day);
#        if($nodateyet) {
#          @mindate = @current;
#          @maxdate = @current;
#          $nodateyet = 0;
#        } else {
#          my $diff1 = Delta_Days(@mindate, @current);
#          if($diff1 < 0) { @mindate = @current; }
#          my $diff2 = Delta_Days(@current, @maxdate);
#          if($diff2 < 0) { @maxdate = @current; }
#
#          #pr2 "$diff1, $diff2";
#        }
#      }
#    }
 

  }

#  my $diff = Delta_Days(@mindate, @maxdate);
#  if($diff == 1 || $diff == 0) {
#    $duration = "1 day";
#  }
#  elsif($diff < 31) {
#    $duration = "$diff days"; 
#  }
#  elsif($diff > 365) {
#    my $years = int($diff / 365);
#    my $s = plural($years);
#    $duration = "$years year$s"; 
#  }
#  elsif($diff > 31) {
#    my $months = int($diff / 31);
#    my $s = plural($months);
#    $duration = "$months month$s"; 
#  }
#  else { die "impossible"; }

  $duration = "xxx months";

  $numauthors = scalar(keys %{$h});
} else {
  pr2 "no GIT INFO?";
}


#------------------------------------------------------------------------------
# Size P (only change for .c in drivers/ or sounds/ (the test files))
#------------------------------------------------------------------------------


foreach my $c (@cfiles) {
  die "wierd: $c, with $spfile" unless ($c =~ /(.*)\.c$/);
  my $base = $1;
  my $bef = "$base.c";
  my $aft = "$base.res"; 
  if(-e "corrected_$base.res") { 
    $aft = "corrected_$base.res";
    mylog "found corrected";
  }
  my $onlychange = 0;
  open TMP, "diff -u -b -B $bef $aft |";
  
  my $count = 0;
  while(<TMP>) {
    $count++;

    if (/^\+[^+]/) { $onlychange++; }
    if (/^\-[^-]/) { $onlychange++; }
  }
  $sumlinePchange += $onlychange;
  $sumlineP2 += $count;
}

#------------------------------------------------------------------------------
# Time
#------------------------------------------------------------------------------
foreach my $c (@cfiles) {
  die "" unless ($c =~ /(.*)\.c$/);
  my $base = $1;
  
  my $diagnosefile = "";
  mylog "$base";

  if(-e "$base.c.ok")        { $ok++; $diagnosefile = "$base.c.ok"; }
  if(-e "$base.c.failed")    { $fa++; $diagnosefile = "$base.c.failed"; }
  if(-e "$base.c.spatch_ok") { $so++; $diagnosefile = "$base.c.spatch_ok"; }
  if(-e "$base.c.gave_up")   { $gu++; $diagnosefile = "$base.c.gave_up"; }

  open TMP, $diagnosefile or die "no diagnose $base: $diagnosefile";
  my $found = 0;
  my $time = 0;
  while(<TMP>) {
# before -test_okfailed
#    if (/real[ \t]+([0-9])+m([0-9]+)[.]([0-9]+)/) {
#      $found++;
#
#      $time = $1 * 60.0;   # minutes
#      $time += $2;         # seconds
#      $time += $3 / 1000.0; # 1/1000th sec.
#
#      pr2 (sprintf "%4.1fs\n", $time);
#      printf "I: %15s  & %4.1fs\n", $c, $time;
#      
#    }
    if (/time: (.*)/) {
      $found++;

      $time = $1;

      mylog (sprintf "%4.1fs\n", $time);
      printf "I: %15s  & %4.1fs\n", $c, $time;
      
    }


  }
  die "not found time information in $diagnosefile"  unless $found == 1;

  $sumtime += $time;
  $maxtime = $time if $time > $maxtime;
  
}

#------------------------------------------------------------------------------
# Computations
#------------------------------------------------------------------------------

my $correct = $ok + $so;

my $pourcentcorrect = ($correct * 100.0) / $nbfiles;
my $avglines = $sumlinefiles / $nbfiles;
my $avgtime = $sumtime / $nbfiles;


my $ratioPvsSP = $sumlineP / $sizeSP;
my $ratioPvsSP2 = $sumlineP2 / $sizeSP;


#------------------------------------------------------------------------------
# Results
#------------------------------------------------------------------------------


pr "SP = $spfile";
mylog "FILES = \n";
map { mylog "\t$_"; } @cfiles;
pr "----------------------------------------";


pr "!!Total files = $nbfiles";
printf "!!AvgLine = %.1fl\n", $avglines;

#pr "  Correct number = $correct";
printf "!!Correct = %.1f%s\n", $pourcentcorrect, "%";

pr "!!Human errors = $errors";

pr "!!Size SP = $sizeSP";
pr "!!Size P = $sumlineP";
pr "!!Size P (change only) = $sumlinePchange";

printf "!!Ratio P/SP = %3.1f\n", $ratioPvsSP;


printf "!!RunTime = %.1fs\n", $sumtime;
printf "!!MaxTime = %.1fs\n", $maxtime;
printf "!!AvgTime = %.1fs\n", $avgtime;

my $totalstatus = $ok + $fa + $so + $gu;
mylog "----------------------------------------------------------------";
mylog "Sanity checks: nb files vs total status: $nbfiles =? $totalstatus";


printf "L: %20s (r%3s) & %5.1f%% & %5dfi & %2de & %6.1fx & %6.1fs \n", 
 $cedescr, $ruleno, $pourcentcorrect, $nbfiles, $errors, $ratioPvsSP, $sumtime;


# Mega, Complex, Bluetooth

printf "M: %60s & %5d & %6d (%d) & %2d & %s & %2d & %3d & %6.0fx & %6.1fs (%.1fs)  & %5.0f\\%% \\\\\\hline%% SP: %s  \n", 
 $cedescr, $nbfiles, $sumlineP, $sumlinePchange, $numauthors, $duration, $errors, $sizeSP, $ratioPvsSP,
 $avgtime, $maxtime, $pourcentcorrect, $spfile;

printf "C: %60s & %5d & %6d (%d) & %2d & %3d & %6.0fx & %6.1fs (%.1fs)  & %5.0f\\%% \\\\\\hline%% SP: %s  \n", 
 $cedescr, $nbfiles, $sumlineP, $sumlinePchange, $errors, $sizeSP, $ratioPvsSP,
 $avgtime, $maxtime, $pourcentcorrect, $spfile;

printf "B: %60s & %5d & %5d (%d) & %3d & %6.0fx & %6.1fs (%.1fs) & %5.0f\\%% \\\\\\hline%% SP: %s  \n", 
 $cedescr, $nbfiles, $sumlineP, $sumlinePchange, $sizeSP, $ratioPvsSP,
 $avgtime, $maxtime, $pourcentcorrect, $spfile;
Commit	Line	Data
34e49164 C	1	#!/usr/bin/perl
	2
	3	use strict;
	4	use diagnostics;
	5
	6	#use Data::Dumper;
	7	#use Date::Manip qw(ParseDate UnixDate); #sudo apt-get install libdate-manip-perl
	8	#use Date::Calc qw(Delta_Days); #sudo apt-get install libdate-calc-perl
	9
	10	#------------------------------------------------------------------------------
	11	# Helpers
	12	#------------------------------------------------------------------------------
	13
	14	my $debug = 0;
	15
	16	sub pr2 { print STDERR "@_\n"; }
	17	sub pr { print "@_\n"; }
	18	sub mylog { print STDERR "@_\n" if $debug; }
	19
	20	sub plural {
	21	my ($n) = @_;
	22	$n > 1 ? "s" : "";
	23	}
	24
	25	#------------------------------------------------------------------------------
	26	# Globals
	27	#------------------------------------------------------------------------------
	28
	29	my $ok = 0;
	30	my $so = 0; #spatch ok
	31	my $fa = 0; #failed
	32	my $gu = 0; #gave up
	33
	34	my $nbfiles = 0;
	35
	36	my $maxtime = 0;
	37	#my $mintime = 0;
	38	my $sumtime = 0;
	39
	40	my $sumlinefiles = 0;
	41
	42	my $errors = 0;
	43
	44	my $sumlineP = 0; #whole git
	45	my $sumlineP2 = 0;
	46	my $sumlinePchange = 0;
	47
	48	my $spfile = "";
	49	my $ruleno = "??";
	50	my $sizeSP = 0;
	51
	52	my $cedescr = "";
	53
	54	my $numauthors = 0;
	55	my $duration = 0; # in days
	56
	57	my @cfiles = ();
	58
	59	#------------------------------------------------------------------------------
	60	# SP
	61	#------------------------------------------------------------------------------
	62	$spfile = `make sp_file`;
	63	chomp $spfile;
	64
65	if($spfile =~ /(rule\|mega\|bt)(\d+)\.cocci/) { $ruleno = "$2"; }
66
67	#------------------------------------------------------------------------------
68	# CE
69	#------------------------------------------------------------------------------
70
71	#$cedescr = `make ce_descr`;
72	#chomp $cedescr;
73	#print STDERR (Dumper($cedescr));
74	open TMP, "Makefile" or die "no Makefile file ?";
75	while(<TMP>) {
76	if(/^(CE)?DESCRIPTION=["'](.*)["']/) { $cedescr = $2; }
77	}
78
79
80
81	#$cedescr =~ s/\\/\\\\/g;
82	#$cedescr =~ s/\f/\\f/g;
83	#$cedescr =~ s/\t/\\t/g;
84
85	#------------------------------------------------------------------------------
86	# List c files
87	#------------------------------------------------------------------------------
88	my $files = `make source_files`;
89	chomp $files;
90	@cfiles = split /\s+/, $files;
91
92	$nbfiles = scalar(@cfiles);
93
94	#------------------------------------------------------------------------------
95	# Size files (lines)
96	#------------------------------------------------------------------------------
97	map {
98	my ($linefile) = `wc -l $_`;
99	chomp $linefile;
100	die "wierd wc output" unless $linefile =~ /^(\d+) /;
101	$sumlinefiles += $1;
102	mylog "filesize $_ $1";
103	} @cfiles;
104
105
106	#------------------------------------------------------------------------------
107	# Size SP
108	#------------------------------------------------------------------------------
109	$sizeSP =
110	`cat $spfile \| perl -p -e "s/\\/\\/.//g;" \| grep -v '^[ \t]\$' \| wc -l`;
111	chomp $sizeSP;
112
113
114	#------------------------------------------------------------------------------
115	# Bugs
116	#------------------------------------------------------------------------------
117	if(!(-e "README")) { pr2 "no README file ?"; }
118	else {
119	open TMP, "README" or die "no README file ?";
120	while(<TMP>) {
121	if (/\[bug\]/ \|\| /status\]\sbug/ \|\| /status\]\sBUG/ ) {
122
123	# can also look for [semibug] but it's often related to [corrected_c] kind of pbs
124	#\|\| /status\]\s*semi-bug/
125
126	#pr2 "OLD BUG FORMAT: $_";
127	$errors++
128	}
129	}
130	}
131
132
133
134	#------------------------------------------------------------------------------
135	# Size P (total)
136	#------------------------------------------------------------------------------
137
138	if(-e "gitinfo") {
139	($sumlineP) = `cat gitinfo \|wc -l`;
140	chomp $sumlineP;
141	} else {
142	pr2 "no GIT INFO?";
143	}
144
145	#------------------------------------------------------------------------------
146	# Number of authors and duration
147	#------------------------------------------------------------------------------
148
149
150	if(-e "gitinfo") {
151
152	open TMP, "gitinfo" or die "no gitinfo file ?";
153
154	#for authors
155	my $h = {};
156
157	#for duration
158	my @mindate = ();
159	my @maxdate = ();
160	my $nodateyet = 1;
161
162	while(<TMP>) {
163	#can also do: egrep "^Author" gitinfo \| sort \| uniq \| wc -l
164	if (/^Author: (.*)/) {
165	$h->{$1}++;
166	}
167
168	# if(/^Date: (.*) ([-+]\d+)?/) {
169	# my $date = ParseDate($1);
170	# if (!$date) { die "bad date" }
171	# else {
172	# my ($year, $month, $day) = UnixDate($date, "%Y", "%m", "%d");
173	# my @current = ($year, $month, $day);
174	# if($nodateyet) {
175	# @mindate = @current;
176	# @maxdate = @current;
177	# $nodateyet = 0;
178	# } else {
179	# my $diff1 = Delta_Days(@mindate, @current);
180	# if($diff1 < 0) { @mindate = @current; }
181	# my $diff2 = Delta_Days(@current, @maxdate);
182	# if($diff2 < 0) { @maxdate = @current; }
183	#
184	# #pr2 "$diff1, $diff2";
185	# }
186	# }
187	# }
188
189
190
191	}
192
193	# my $diff = Delta_Days(@mindate, @maxdate);
194	# if($diff == 1 \|\| $diff == 0) {
195	# $duration = "1 day";
196	# }
197	# elsif($diff < 31) {
198	# $duration = "$diff days";
199	# }
200	# elsif($diff > 365) {
201	# my $years = int($diff / 365);
202	# my $s = plural($years);
203	# $duration = "$years year$s";
204	# }
205	# elsif($diff > 31) {
206	# my $months = int($diff / 31);
207	# my $s = plural($months);
208	# $duration = "$months month$s";
209	# }
210	# else { die "impossible"; }
211
212	$duration = "xxx months";
213
214	$numauthors = scalar(keys %{$h});
215	} else {
216	pr2 "no GIT INFO?";
217	}
218
219
220	#------------------------------------------------------------------------------
221	# Size P (only change for .c in drivers/ or sounds/ (the test files))
222	#------------------------------------------------------------------------------
223
224
225	foreach my $c (@cfiles) {
226	die "wierd: $c, with $spfile" unless ($c =~ /(.*)\.c$/);
227	my $base = $1;
228	my $bef = "$base.c";
229	my $aft = "$base.res";
230	if(-e "corrected_$base.res") {
231	$aft = "corrected_$base.res";
232	mylog "found corrected";
233	}
234	my $onlychange = 0;
235	open TMP, "diff -u -b -B $bef $aft \|";
236
237	my $count = 0;
238	while(<TMP>) {
239	$count++;
240
241	if (/^\+[^+]/) { $onlychange++; }
242	if (/^\-[^-]/) { $onlychange++; }
243	}
244	$sumlinePchange += $onlychange;
245	$sumlineP2 += $count;
246	}
247
248	#------------------------------------------------------------------------------
249	# Time
250	#------------------------------------------------------------------------------
251	foreach my $c (@cfiles) {
252	die "" unless ($c =~ /(.*)\.c$/);
253	my $base = $1;
254
255	my $diagnosefile = "";
256	mylog "$base";
257
258	if(-e "$base.c.ok") { $ok++; $diagnosefile = "$base.c.ok"; }
259	if(-e "$base.c.failed") { $fa++; $diagnosefile = "$base.c.failed"; }
260	if(-e "$base.c.spatch_ok") { $so++; $diagnosefile = "$base.c.spatch_ok"; }
261	if(-e "$base.c.gave_up") { $gu++; $diagnosefile = "$base.c.gave_up"; }
262
263	open TMP, $diagnosefile or die "no diagnose $base: $diagnosefile";
264	my $found = 0;
265	my $time = 0;
266	while(<TMP>) {
267	# before -test_okfailed
268	# if (/real[ \t]+([0-9])+m([0-9]+)[.]([0-9]+)/) {
269	# $found++;
270	#
271	# $time = $1 * 60.0; # minutes
272	# $time += $2; # seconds
273	# $time += $3 / 1000.0; # 1/1000th sec.
274	#
275	# pr2 (sprintf "%4.1fs\n", $time);
276	# printf "I: %15s & %4.1fs\n", $c, $time;
277	#
278	# }
279	if (/time: (.*)/) {
280	$found++;
281
282	$time = $1;
283
284	mylog (sprintf "%4.1fs\n", $time);
285	printf "I: %15s & %4.1fs\n", $c, $time;
286
287	}
288
289
290	}
291	die "not found time information in $diagnosefile" unless $found == 1;
292
293	$sumtime += $time;
294	$maxtime = $time if $time > $maxtime;
295
296	}
297
298	#------------------------------------------------------------------------------
299	# Computations
300	#------------------------------------------------------------------------------
301
302	my $correct = $ok + $so;
303
304	my $pourcentcorrect = ($correct * 100.0) / $nbfiles;
305	my $avglines = $sumlinefiles / $nbfiles;
306	my $avgtime = $sumtime / $nbfiles;
307
308
309	my $ratioPvsSP = $sumlineP / $sizeSP;
310	my $ratioPvsSP2 = $sumlineP2 / $sizeSP;
311
312
313	#------------------------------------------------------------------------------
314	# Results
315	#------------------------------------------------------------------------------
316
317
318	pr "SP = $spfile";
319	mylog "FILES = \n";
320	map { mylog "\t$_"; } @cfiles;
321	pr "----------------------------------------";
322
323
324	pr "!!Total files = $nbfiles";
325	printf "!!AvgLine = %.1fl\n", $avglines;
326
327	#pr " Correct number = $correct";
328	printf "!!Correct = %.1f%s\n", $pourcentcorrect, "%";
329
330	pr "!!Human errors = $errors";
331
332	pr "!!Size SP = $sizeSP";
333	pr "!!Size P = $sumlineP";
334	pr "!!Size P (change only) = $sumlinePchange";
335
336	printf "!!Ratio P/SP = %3.1f\n", $ratioPvsSP;
337
338
339	printf "!!RunTime = %.1fs\n", $sumtime;
340	printf "!!MaxTime = %.1fs\n", $maxtime;
341	printf "!!AvgTime = %.1fs\n", $avgtime;
342
343	my $totalstatus = $ok + $fa + $so + $gu;
344	mylog "----------------------------------------------------------------";
345	mylog "Sanity checks: nb files vs total status: $nbfiles =? $totalstatus";
346
347
348
349	printf "L: %20s (r%3s) & %5.1f%% & %5dfi & %2de & %6.1fx & %6.1fs \n",
350	$cedescr, $ruleno, $pourcentcorrect, $nbfiles, $errors, $ratioPvsSP, $sumtime;
351
352
353	# Mega, Complex, Bluetooth
354
355	printf "M: %60s & %5d & %6d (%d) & %2d & %s & %2d & %3d & %6.0fx & %6.1fs (%.1fs) & %5.0f\\%% \\\\\\hline%% SP: %s \n",
356	$cedescr, $nbfiles, $sumlineP, $sumlinePchange, $numauthors, $duration, $errors, $sizeSP, $ratioPvsSP,
357	$avgtime, $maxtime, $pourcentcorrect, $spfile;
358
359	printf "C: %60s & %5d & %6d (%d) & %2d & %3d & %6.0fx & %6.1fs (%.1fs) & %5.0f\\%% \\\\\\hline%% SP: %s \n",
360	$cedescr, $nbfiles, $sumlineP, $sumlinePchange, $errors, $sizeSP, $ratioPvsSP,
361	$avgtime, $maxtime, $pourcentcorrect, $spfile;
362
363	printf "B: %60s & %5d & %5d (%d) & %3d & %6.0fx & %6.1fs (%.1fs) & %5.0f\\%% \\\\\\hline%% SP: %s \n",
364	$cedescr, $nbfiles, $sumlineP, $sumlinePchange, $sizeSP, $ratioPvsSP,
365	$avgtime, $maxtime, $pourcentcorrect, $spfile;
366
367