florian | b0f4ce8 | 2011-10-25 20:01:41 +0000 | [diff] [blame] | 1 | #!/usr/bin/env perl |
| 2 | |
| 3 | #----------------------------------------------------------------- |
| 4 | # Quick and dirty script to summarize build information for a |
| 5 | # set of nightly runs. |
| 6 | # |
| 7 | # The results of the nighly regression runs are extracted from |
| 8 | # the GMANE mail archive. The URL for a given mail sent to the |
| 9 | # valgrind-developers mailing list is |
| 10 | # |
| 11 | # http://article.gmane.org/gmane.comp.debugging.valgrind.devel/<integer> |
| 12 | # |
| 13 | # The script extracts information about the regression run from a |
| 14 | # block of information at the beginning of the mail. That information |
| 15 | # was added beginning October 4, 2011. Therefore, only regression runs |
| 16 | # from that date or later can be analyzed. |
| 17 | # |
| 18 | # There is unfortunately no good way of figuring out the interval |
| 19 | # of integers in the above URL that include all nightly regression |
| 20 | # runs. |
| 21 | # |
| 22 | # The function get_regtest_data does all the work. It returns a hash |
| 23 | # whose keys are the dates at which nightly runs took place. The value |
| 24 | # is in turn a hash. |
| 25 | # |
| 26 | # Each such hash has the following keys: |
| 27 | # "builds" array of hashes |
| 28 | # "num_builds" int |
| 29 | # "num_failing_builds" int |
| 30 | # "num_passing_builds" int |
| 31 | # "num_testcase_failures" int |
| 32 | # "num_failing_testcases" int |
| 33 | # "failure_frequency" hash indexed by testcase name; value = int |
| 34 | # |
| 35 | # "builds" is an array of hashes with the following keys |
| 36 | # "arch" string (architecture) |
| 37 | # "distro" string (distribution, e.g. Fedora-15) |
| 38 | # "failures" array of strings (failing testcases) |
| 39 | # "valgrind revision" integer |
| 40 | # "VEX revision" integer |
| 41 | # "GCC version" string |
| 42 | # "C library" string |
| 43 | # "uname -mrs" string |
| 44 | # "Vendor version" string |
| 45 | # |
| 46 | #----------------------------------------------------------------- |
| 47 | use strict; |
| 48 | use warnings; |
| 49 | |
| 50 | use LWP::Simple; |
| 51 | use Getopt::Long; |
| 52 | |
| 53 | my $prog_name = "nightly-build-summary"; |
| 54 | |
| 55 | my $debug = 0; |
| 56 | my $keep = 0; |
| 57 | |
| 58 | my $usage=<<EOF; |
| 59 | USAGE |
| 60 | |
| 61 | $prog_name |
| 62 | |
| 63 | --from=INTEGER beginning of mail interval; > 14800 |
| 64 | |
| 65 | [--to=INTEGER] end of mail interval; default = from + 100 |
| 66 | |
| 67 | [--debug] verbose mode (debugging) |
| 68 | |
| 69 | [--keep] write individual emails to files (debugging) |
| 70 | |
| 71 | [--dump] write results suitable for post-processing |
| 72 | |
| 73 | [--readable] write results in human readable form (default) |
| 74 | |
| 75 | EOF |
| 76 | |
| 77 | |
| 78 | #----------------------------------------------------------------- |
| 79 | # Search for a line indicating that this is an email containing |
| 80 | # the results of a valgrind regression run. |
| 81 | # Return 1, if found and 0 oherwise. |
| 82 | #----------------------------------------------------------------- |
| 83 | sub is_regtest_result { |
| 84 | my (@lines) = @_; |
| 85 | |
| 86 | foreach my $line (@lines) { |
| 87 | return 1 if ($line =~ "^valgrind revision:"); |
| 88 | } |
| 89 | |
| 90 | return 0; |
| 91 | } |
| 92 | |
| 93 | |
| 94 | #----------------------------------------------------------------- |
| 95 | # Extract information from the run. Don't prep the data here. This |
| 96 | # is done later on. |
| 97 | #----------------------------------------------------------------- |
| 98 | sub get_raw_data { |
| 99 | my (@lines, $msgno) = @_; |
| 100 | my ($i, $n, $line, $date); |
| 101 | |
| 102 | $n = scalar @lines; |
| 103 | |
| 104 | my %hash = (); |
| 105 | |
florian | d8e32e2 | 2011-10-28 02:45:15 +0000 | [diff] [blame] | 106 | # 1) Locate the section with the info about the environment of this nightly run |
florian | d1e996a | 2013-09-10 14:00:45 +0000 | [diff] [blame^] | 107 | for ($i = 0; $i < $n; ++$i) { |
florian | b0f4ce8 | 2011-10-25 20:01:41 +0000 | [diff] [blame] | 108 | last if ($lines[$i] =~ /^valgrind revision:/); |
| 109 | } |
| 110 | die "no info block in message $msgno" if ($i == $n); |
| 111 | |
florian | d8e32e2 | 2011-10-28 02:45:15 +0000 | [diff] [blame] | 112 | # 2) Read the info about the build: compiler, valgrind revision etc. |
florian | b0f4ce8 | 2011-10-25 20:01:41 +0000 | [diff] [blame] | 113 | # and put it into a hash. |
| 114 | for ( ; $i < $n; ++$i) { |
| 115 | $line = $lines[$i]; |
| 116 | last if ($line =~ /^$/); # empty line indicates end of section |
| 117 | my ($key, $value) = split(/:/, $line); |
| 118 | $value =~ s/^[ ]*//; # removing leading blanks |
| 119 | $hash{$key} = $value; |
| 120 | } |
| 121 | |
| 122 | if ($debug) { |
| 123 | foreach my $key (keys %hash) { |
| 124 | my ($val) = $hash{$key}; |
| 125 | print "regtest env: KEY = |$key| VAL = |$val|\n"; |
| 126 | } |
| 127 | } |
| 128 | |
florian | d8e32e2 | 2011-10-28 02:45:15 +0000 | [diff] [blame] | 129 | # 3) Get the date from when the build was kicked off. |
| 130 | for ( ; $i < $n; ++$i) { |
| 131 | $line = $lines[$i]; |
| 132 | |
| 133 | if ($line =~ /^Started at[ ]+([^ ]+)/) { |
| 134 | $date = $1; |
| 135 | print "DATE = $date\n"; |
| 136 | last; |
| 137 | } |
| 138 | } |
| 139 | die "no date found in message $msgno" if ($i == $n); |
| 140 | |
| 141 | |
florian | b0f4ce8 | 2011-10-25 20:01:41 +0000 | [diff] [blame] | 142 | # 4) Find out if the regression run failed or passed |
| 143 | $hash{"failures"} = []; |
| 144 | for ($i = $i + 1; $i < $n; ++$i) { |
| 145 | $line = $lines[$i]; |
| 146 | if ($line =~ /Running regression tests/) { |
| 147 | return %hash if ($line =~ /done$/); # regtest succeeded; no failures |
| 148 | die "cannot determine regtest outcome for message $msgno" |
| 149 | if (! ($line =~ /failed$/)); |
| 150 | last; |
| 151 | } |
| 152 | } |
| 153 | |
| 154 | # 5) Regtest failed; locate the section with the list of failing testcases |
| 155 | for ($i = $i + 1; $i < $n; ++$i) { |
| 156 | $line = $lines[$i]; |
| 157 | # Match for end-of-line == because line might be split. |
| 158 | last if ($line =~ /==$/); |
| 159 | } |
| 160 | die "cannot locate failing testcases in message $msgno" if ($i == $n); |
| 161 | |
| 162 | # 6) Get list of failing testcases |
| 163 | for ($i = $i + 1; $i < $n; ++$i) { |
| 164 | $line = $lines[$i]; |
| 165 | |
| 166 | last if ($line =~ /^$/); |
| 167 | |
| 168 | my ($testcase) = (split(/\s+/, $line))[0]; |
| 169 | print "ADD failing testcase $testcase\n" if ($debug); |
| 170 | push @{$hash{"failures"}}, $testcase; |
| 171 | } |
| 172 | |
| 173 | return ($date, %hash); |
| 174 | } |
| 175 | |
| 176 | |
| 177 | #----------------------------------------------------------------- |
| 178 | # Extract architecture; get a pretty name for the distro |
| 179 | #----------------------------------------------------------------- |
| 180 | sub prep_regtest_data { |
| 181 | my (%hash) = @_; |
| 182 | my ($val, $arch, $distro); |
| 183 | |
| 184 | $val = $hash{"uname -mrs"}; |
| 185 | die "uname -mrs info is missing" if (! defined $val); |
| 186 | $arch = (split(/ /, $val))[2]; |
| 187 | |
| 188 | $val = $hash{"Vendor version"}; |
| 189 | die "Vendor version info is missing" if (! defined $val); |
| 190 | |
| 191 | if ($val =~ /Fedora release ([0-9]+)/) { |
| 192 | $distro = "Fedora-$1"; |
| 193 | } elsif ($val =~ /openSUSE ([0-9]+)\.([0-9]+)/) { |
| 194 | $distro = "openSUSE-$1.$2"; |
| 195 | } elsif ($val =~ /SUSE Linux Enterprise Server 11 SP1/) { |
| 196 | $distro = "SLES-11-SP1"; |
| 197 | } elsif ($val =~ /Red Hat Enterprise Linux AS release 4/) { |
| 198 | $distro = "RHEL-4"; |
| 199 | } else { |
| 200 | $distro = "UNKNOWN"; |
| 201 | } |
| 202 | |
| 203 | # Add architecture and distribution to hash |
| 204 | $hash{"arch"} = $arch; |
| 205 | $hash{"distro"} = $distro; |
| 206 | |
| 207 | return %hash; |
| 208 | } |
| 209 | |
| 210 | |
| 211 | #----------------------------------------------------------------- |
| 212 | # Precompute some summary information and record it |
| 213 | #----------------------------------------------------------------- |
| 214 | sub precompute_summary_info |
| 215 | { |
| 216 | my (%dates) = @_; |
| 217 | |
| 218 | foreach my $date (sort keys %dates) { |
| 219 | my %failure_frequency = (); |
| 220 | |
| 221 | my %nightly = %{ $dates{$date} }; |
| 222 | my @builds = @{ $nightly{"builds"} }; |
| 223 | |
| 224 | $nightly{"num_builds"} = scalar (@builds); |
| 225 | $nightly{"num_failing_builds"} = 0; |
| 226 | $nightly{"num_testcase_failures"} = 0; |
| 227 | |
| 228 | foreach my $build (@builds) { |
| 229 | my %regtest_data = %{ $build }; |
| 230 | |
| 231 | my @failures = @{ $regtest_data{"failures"} }; |
| 232 | my $num_fail = scalar (@failures); |
| 233 | |
| 234 | ++$nightly{"num_failing_builds"} if ($num_fail != 0); |
| 235 | $nightly{"num_testcase_failures"} += $num_fail; |
| 236 | |
| 237 | # Compute how often a testcase failed |
| 238 | foreach my $test ( @failures ) { |
| 239 | if (defined $failure_frequency{$test}) { |
| 240 | ++$failure_frequency{$test}; |
| 241 | } else { |
| 242 | $failure_frequency{$test} = 1; |
| 243 | } |
| 244 | } |
| 245 | } |
| 246 | |
| 247 | $nightly{"num_passing_builds"} = |
| 248 | $nightly{"num_builds"} - $nightly{"num_failing_builds"}; |
| 249 | |
| 250 | $nightly{"num_failing_testcases"} = scalar (keys %failure_frequency); |
| 251 | |
| 252 | $nightly{"failure_frequency"} = { %failure_frequency }; |
| 253 | |
| 254 | $dates{$date} = { %nightly }; |
| 255 | } |
| 256 | |
| 257 | return %dates; |
| 258 | } |
| 259 | |
| 260 | |
| 261 | #----------------------------------------------------------------- |
| 262 | # Get messages from GMANE, and build up a database of results. |
| 263 | #----------------------------------------------------------------- |
| 264 | sub get_regtest_data { |
| 265 | my ($from, $to) = @_; |
| 266 | |
| 267 | my $url_base = "http://article.gmane.org/gmane.comp.debugging.valgrind.devel/"; |
| 268 | |
| 269 | my %dates = (); |
| 270 | |
| 271 | my $old_date = "-1"; |
| 272 | my @builds = (); |
| 273 | |
| 274 | for (my $i = $from; $i <= $to; ++$i) { |
| 275 | my $url = "$url_base" . "$i"; |
| 276 | |
| 277 | my $page = get("$url"); |
| 278 | |
| 279 | if ($keep) { |
| 280 | open (EMAIL, ">$i"); |
| 281 | print EMAIL $page; |
| 282 | close(EMAIL); |
| 283 | } |
| 284 | |
| 285 | # Detect if the article does not exist. Happens for too large --to= values |
| 286 | last if ($page eq "No such file.\n"); |
| 287 | |
| 288 | # Split the page into lines |
| 289 | my @lines = split(/\n/, $page); |
| 290 | |
| 291 | # Check whether it contains a regression test result |
| 292 | next if (! is_regtest_result(@lines)); |
| 293 | print "message $i is a regression test result\n" if ($debug); |
| 294 | |
| 295 | # Get the raw data |
| 296 | my ($date, %regtest_data) = get_raw_data(@lines); |
| 297 | |
| 298 | %regtest_data = prep_regtest_data(%regtest_data); |
| 299 | |
| 300 | if ($date ne $old_date) { |
| 301 | my %nightly = (); |
| 302 | $nightly{"builds"} = [ @builds ]; |
| 303 | $dates{$old_date} = { %nightly } if ($old_date ne "-1"); |
| 304 | |
| 305 | $old_date = $date; |
| 306 | @builds = (); |
| 307 | } |
| 308 | |
| 309 | push @builds, { %regtest_data }; |
| 310 | } |
| 311 | my %nightly = (); |
| 312 | $nightly{"builds"} = [ @builds ]; |
| 313 | $dates{$old_date} = { %nightly } if ($old_date ne "-1"); |
| 314 | |
| 315 | # Convenience: precompute some info we'll be interested in |
| 316 | %dates = precompute_summary_info( %dates ); |
| 317 | |
| 318 | return %dates; |
| 319 | } |
| 320 | |
| 321 | |
| 322 | #----------------------------------------------------------------- |
| 323 | # Write out the results in a form suitable for automatic post-processing |
| 324 | #----------------------------------------------------------------- |
| 325 | sub dump_results { |
| 326 | my (%dates) = @_; |
| 327 | |
| 328 | foreach my $date (sort keys %dates) { |
| 329 | |
| 330 | my %nightly = %{ $dates{$date} }; |
| 331 | my @builds = @{ $nightly{"builds"} }; |
| 332 | |
| 333 | foreach my $build (@builds) { |
| 334 | my %regtest_data = %{ $build }; |
| 335 | |
| 336 | my $arch = $regtest_data{"arch"}; |
| 337 | my $distro = $regtest_data{"distro"}; |
| 338 | my @failures = @{ $regtest_data{"failures"} }; |
| 339 | my $num_fail = scalar (@failures); |
| 340 | my $fails = join(":", sort @failures); |
| 341 | |
| 342 | printf("Regrun: %s %3d %-10s %-20s %s\n", |
| 343 | $date, $num_fail, $arch, $distro, $fails); |
| 344 | } |
| 345 | |
| 346 | my %failure_frequency = %{ $nightly{"failure_frequency"} }; |
| 347 | |
| 348 | foreach my $test (keys %failure_frequency) { |
| 349 | printf("Test: %s %3d %s\n", |
| 350 | $date, $failure_frequency{$test}, $test); |
| 351 | } |
| 352 | |
| 353 | printf("Total: %s builds: %d %d fail %d pass tests: %d fail %d unique\n", |
| 354 | $date, $nightly{"num_builds"}, $nightly{"num_failing_builds"}, |
| 355 | $nightly{"num_passing_builds"}, $nightly{"num_testcase_failures"}, |
| 356 | $nightly{"num_failing_testcases"}); |
| 357 | } |
| 358 | } |
| 359 | |
| 360 | |
| 361 | sub write_readable_results { |
| 362 | my (%dates) = @_; |
| 363 | |
| 364 | foreach my $date (sort keys %dates) { |
| 365 | my %nightly = %{ $dates{$date} }; |
| 366 | |
| 367 | print "$date\n----------\n"; |
| 368 | |
| 369 | printf("%3d builds\n", $nightly{"num_builds"}); |
| 370 | printf("%3d builds fail\n", $nightly{"num_failing_builds"}); |
| 371 | printf("%3d builds pass\n", $nightly{"num_passing_builds"}); |
| 372 | print "\n"; |
| 373 | printf("%3d testcase failures (across all runs)\n", |
| 374 | $nightly{"num_testcase_failures"}); |
| 375 | printf("%3d failing testcases (unique)\n", |
| 376 | $nightly{"num_failing_testcases"}); |
| 377 | print "\n"; |
| 378 | |
| 379 | my @builds = @{ $nightly{"builds"} }; |
| 380 | |
| 381 | if ($nightly{"num_passing_builds"} != 0) { |
| 382 | print "Passing builds\n"; |
| 383 | print "--------------\n"; |
| 384 | foreach my $build (@builds) { |
| 385 | my %regtest_data = %{ $build }; |
| 386 | my @failures = @{ $regtest_data{"failures"} }; |
| 387 | my $num_fail = scalar (@failures); |
| 388 | |
| 389 | if ($num_fail == 0) { |
| 390 | my $arch = $regtest_data{"arch"}; |
| 391 | my $distro = $regtest_data{"distro"}; |
| 392 | |
| 393 | printf("%-8s %-15s\n", $arch, $distro); |
| 394 | } |
| 395 | print "\n"; |
| 396 | } |
| 397 | print "\n"; |
| 398 | } |
| 399 | |
| 400 | if ($nightly{"num_failing_builds"} != 0) { |
| 401 | print "Failing builds\n"; |
| 402 | print "--------------\n"; |
| 403 | foreach my $build (@builds) { |
| 404 | my %regtest_data = %{ $build }; |
| 405 | my @failures = @{ $regtest_data{"failures"} }; |
| 406 | my $num_fail = scalar (@failures); |
| 407 | |
| 408 | if ($num_fail != 0) { |
| 409 | my $arch = $regtest_data{"arch"}; |
| 410 | my $distro = $regtest_data{"distro"}; |
| 411 | |
| 412 | printf("%-8s %-15s %d failures\n", $arch, $distro, $num_fail); |
| 413 | foreach my $test (@failures) { |
| 414 | print " $test\n"; |
| 415 | } |
| 416 | print "\n"; |
| 417 | } |
| 418 | } |
| 419 | print "\n"; |
| 420 | } |
| 421 | |
| 422 | print "Failing testcases and their frequency\n"; |
| 423 | print "-------------------------------------\n"; |
| 424 | my %failure_frequency = %{ $nightly{"failure_frequency"} }; |
| 425 | |
| 426 | # Sorted in decreasing frequency |
| 427 | foreach my $test (sort {$failure_frequency{$b} cmp $failure_frequency{$a} } |
| 428 | keys %failure_frequency) { |
| 429 | printf("%3d %s\n", $failure_frequency{$test}, $test); |
| 430 | } |
| 431 | print "\n"; |
| 432 | } |
| 433 | } |
| 434 | |
| 435 | |
| 436 | sub main |
| 437 | { |
| 438 | my ($from, $to, $dump, $readable); |
| 439 | |
| 440 | $from = $to = 0; |
| 441 | $dump = $readable = 0; |
| 442 | |
| 443 | GetOptions( "from=i" => \$from, |
| 444 | "to=i" => \$to, |
| 445 | "debug" => \$debug, |
| 446 | "dump" => \$dump, |
| 447 | "keep" => \$keep, |
| 448 | "readable" => \$readable |
| 449 | ) || die $usage; |
| 450 | |
| 451 | # 14800 is about Oct 4, 2011 which is when we began including information |
| 452 | # about the environment |
| 453 | |
| 454 | die $usage if ($from < 14800); |
| 455 | |
| 456 | $to = $from + 100 if ($to == 0); |
| 457 | |
| 458 | if ($from > $to) { |
| 459 | print STDERR "*** invalid [from,to] interval. Try again\n"; |
| 460 | die $usage; |
| 461 | } |
| 462 | |
| 463 | $readable = 1 if ($dump == 0 && $readable == 0); |
| 464 | |
| 465 | print "check message interval [$from...$to]\n" if ($debug); |
| 466 | |
| 467 | # Get mails from GMANE mail archive |
| 468 | |
| 469 | my %dates = get_regtest_data($from, $to); |
| 470 | |
| 471 | dump_results(%dates) if ($dump); |
| 472 | |
| 473 | write_readable_results(%dates) if ($readable); |
| 474 | } |
| 475 | |
| 476 | main(); |
| 477 | |
| 478 | exit 0; |