bin/lo-commit-stat

   1 #!/usr/bin/perl
   2     eval 'exec /usr/bin/perl -S $0 ${1+"$@"}'
   3         if $running_under_some_shell;
   4 #!/usr/bin/perl
   5
   6 use strict;
   7 use LWP::UserAgent;
   8 use utf8;
   9 use File::Temp;
  10
  11 my %module_dirname = (
  12     "core"              => "",
  13     "dictionaries"      => "dictionaries",
  14     "help"              => "helpcontent2",
  15     "translations"      => "translations"
  16 );
  17
  18
  19 my %bugzillas = (
  20     fdo  => "https://bugs.freedesktop.org/show_bug.cgi?id=",
  21     bnc  => "https://bugzilla.novell.com/show_bug.cgi?id=",
  22     rhbz => "https://bugzilla.redhat.com/show_bug.cgi?id=",
  23     i    => "https://issues.apache.org/ooo/show_bug.cgi?id=",
  24     fate => "https://features.opensuse.org/",
  25 );
  26
  27 sub search_bugs($$$$)
  28 {
  29     my ($pdata, $module, $commit_id, $line) = @_;
  30
  31     my $bug = "";
  32     my $bug_orig;
  33     while (defined $bug) {
  34
  35         # match fdo#123, rhz#123, i#123, #123
  36         # but match only bug number with >= 4 digits
  37         if ( $line =~ m/(\w+\#+\d{4,})/ ) {
  38             $bug_orig = $1;
  39             $bug = $1;
  40                 # default to issuezilla for the #123 variant
  41         # but match only bug number with >= 4 digits
  42         } elsif ( $line =~ m/(\#)(\d{4,})/ ) {
  43             $bug_orig = $1 . $2;
  44             $bug = "i#$2";
  45         # match #i123#
  46         } elsif ( $line =~ m/(\#i)(\d+)(\#)/ ) {
  47             $bug_orig = $1 . $2 . $3;
  48             $bug = "i#$2";
  49         } else {
  50             $bug = undef;
  51             next;
  52         }
  53
  54 #        print "   found $bug\n";
  55         # remove bug number from the comment; it will be added later a standardized way
  56         $bug_orig =~ s/\#/\\#/;
  57         $line =~ s/[Rr]esolve[ds]:?\s*$bug_orig\s*//;
  58         $line =~ s/\s*-\s*$bug_orig\s*//;
  59         $line =~ s/\(?$bug_orig\)?\s*[:,-]?\s*//;
  60
  61         # bnc# is preferred over n# for novell bugs
  62         $bug =~ s/^n\#/bnc#/;
  63         # deb# is preferred over debian# for debian bugs
  64         $bug =~ s/^debian\#/deb#/;
  65         # easyhack# is sometimes used for fdo# - based easy hacks
  66         $bug =~ s/^easyhack\#/fdo#/;
  67         # someone mistyped fdo as fd0
  68         $bug =~ s/^fd0\#/fdo#/;
  69         # save the bug number
  70         %{$pdata->{$module}{$commit_id}{'bugs'}} = () if (! defined %{$pdata->{$module}{$commit_id}{'bugs'}});
  71         $pdata->{$module}{$commit_id}{'bugs'}{$bug} = 1;
  72     }
  73
  74     return $line;
  75 }
  76
  77 sub standardize_summary($)
  78 {
  79     my $line = shift;
  80
  81     $line =~ s/^\s*//;
  82     $line =~ s/\s*$//;
  83
  84     # lower first letter if the word contains only lowercase letter
  85     if ( $line =~ m/(^.[a-z]+\b)/ ) {
  86         $line =~ m/(^.)/;
  87         my $first_char = lc($1);
  88         $line =~ s/^./$first_char/;
  89     }
  90
  91     # FIXME: remove do at the end of line
  92     #        remove bug numbers
  93     return $line;
  94 }
  95
  96 sub generate_git_cherry_ids_log($$$$$)
  97 {
  98     my ($pdata, $repo_dir, $module, $branch_name, $git_args) = @_;
  99
 100     my $commit_ids_log;
 101     my $commit_ids_log_fh;
 102     $commit_ids_log_fh = File::Temp->new(TEMPLATE => 'lo-commit-stat-ids-XXXXXX',
 103                                          DIR => '/tmp',
 104                                          UNLINK => 0);
 105     $commit_ids_log = $commit_ids_log_fh->filename;
 106
 107     print STDERR "Filtering cherry-picked commits in the git repo: $module...\n";
 108
 109     my $cmd = "cd $repo_dir; git cherry $git_args";
 110     open (GIT, "$cmd 2>&1|") || die "Can't run $cmd: $!";
 111
 112     while (my $line = <GIT>) {
 113
 114         # skip cherry-picked commits
 115         next if ( $line =~ m/^\-/ );
 116
 117         if ( $line =~ m/^\+ / ) {
 118             $line =~ s/^\+ //;
 119             print $commit_ids_log_fh $line;
 120         }
 121     }
 122
 123     close GIT;
 124     close $commit_ids_log_fh;
 125
 126     return $commit_ids_log;
 127 }
 128
 129 sub load_git_log($$$$$$$)
 130 {
 131     my ($pdata, $repo_dir, $module, $branch_name, $git_command, $git_cherry, $git_args) = @_;
 132
 133     my $cmd = "cd $repo_dir;";
 134     my $commit_ids_log;
 135
 136     if ($git_cherry) {
 137         $commit_ids_log = generate_git_cherry_ids_log($pdata, $repo_dir, $module, $branch_name, $git_args);
 138         $cmd .= " cat $commit_ids_log | xargs -n 1 $git_command -1";
 139     } else {
 140         $cmd .= " $git_command $git_args";
 141     }
 142
 143     my $commit_id;
 144     my $summary;
 145
 146     print STDERR "Analyzing log from the git repo: $module...\n";
 147
 148 #   FIXME: ./g pull move submodules in unnamed branches
 149 #    my $repo_branch_name = get_branch_name($repo_dir);
 150 #    if ( $branch_name ne $repo_branch_name ) {
 151 #        die "Error: mismatch of branches:\n" .
 152 #            "       main repo is on the branch: $branch_name\n" .
 153 #            "       $module repo is on the branch: $repo_branch_name\n";
 154 #    }
 155
 156     open (GIT, "$cmd 2>&1|") || die "Can't run $cmd: $!";
 157     %{$pdata->{$module}} = ();
 158
 159     while (my $line = <GIT>) {
 160         chomp $line;
 161
 162         if ( $line =~ m/^commit ([0-9a-z]{20})/ ) {
 163             $commit_id = "$1";
 164             $summary=undef;
 165             %{$pdata->{$module}{"$commit_id"}} = ();
 166             next;
 167         }
 168
 169         if ( $line =~ /^Author:\s*([^\<]*)\<([^\>]*)>/ ) {
 170             # get rid of extra empty spaces;
 171             my $name = "$1";
 172             $name =~ s/\s+$//;
 173             die "Error: Author already defined for the commit {$commit_id}\n" if defined ($pdata->{$module}{$commit_id}{'author'});
 174             %{$pdata->{$module}{$commit_id}{'author'}} = ();
 175             $pdata->{$module}{$commit_id}{'author'}{'name'} = "$name";
 176             $pdata->{$module}{$commit_id}{'author'}{'email'} = "$2";
 177             next;
 178         }
 179
 180         if ( $line =~ /^Date:\s+/ ) {
 181             # ignore date line
 182             next;
 183         }
 184
 185         if ( $line =~ /^\s*$/ ) {
 186             # ignore empty line
 187             next;
 188         }
 189
 190         $line = search_bugs($pdata, $module, $commit_id, $line);
 191         # FIXME: need to be implemented
 192 #        search_keywords($pdata, $line);
 193
 194         unless (defined $pdata->{$module}{$commit_id}{'summary'}) {
 195             $summary = standardize_summary($line);
 196             $pdata->{$module}{$commit_id}{'summary'} = $summary;
 197         }
 198     }
 199
 200     close GIT;
 201     unlink $commit_ids_log if ($git_cherry);
 202 }
 203
 204 sub get_repo_name($)
 205 {
 206     my $repo_dir = shift;
 207
 208     open (GIT_CONFIG, "$repo_dir/.git/config") ||
 209         die "can't open \"$$repo_dir/.git/config\" for reading: $!\n";
 210
 211     while (my $line = <GIT_CONFIG>) {
 212         chomp $line;
 213
 214         if ( $line =~ /^\s*url\s*=\s*(\S+)$/ ) {
 215             my $repo_name = "$1";
 216             $repo_name = s/.*\///g;
 217             return "$repo_name";
 218         }
 219     }
 220     die "Error: can't find repo name in \"$$repo_dir/.git/config\"\n";
 221 }
 222
 223 sub load_data($$$$$$$)
 224 {
 225     my ($pdata, $top_dir, $p_module_dirname, $branch_name, $git_command, $git_cherry, $git_args) = @_;
 226
 227     foreach my $module (sort { $a cmp $b } keys %{$p_module_dirname}) {
 228         load_git_log($pdata, "$top_dir/$p_module_dirname->{$module}", $module, $branch_name, $git_command, $git_cherry, $git_args);
 229     }
 230 }
 231
 232 sub get_branch_name($)
 233 {
 234     my ($top_dir) = @_;
 235
 236     my $branch;
 237     my $cmd = "cd $top_dir && git branch";
 238
 239     open (GIT, "$cmd 2>&1|") || die "Can't run $cmd: $!";
 240
 241     while (my $line = <GIT>) {
 242         chomp $line;
 243
 244         if ( $line =~ m/^\*\s*(\S+)/ ) {
 245             $branch = "$1";
 246         }
 247     }
 248
 249     close GIT;
 250
 251     die "Error: did not detect git branch name in $top_dir\n" unless defined ($branch);
 252
 253     return $branch;
 254 }
 255
 256 sub get_bug_list($$$)
 257 {
 258     my ($pdata, $pbugs, $check_bugzilla) = @_;
 259
 260     # associate bugs with their summaries and fixers
 261     foreach my $module ( keys %{$pdata}) {
 262         foreach my $id ( keys %{$pdata->{$module}}) {
 263             foreach my $bug (keys %{$pdata->{$module}{$id}{'bugs'}}) {
 264                 %{$pbugs->{$bug}} = () if (! defined %{$pbugs->{$bug}});
 265                 my $author = $pdata->{$module}{$id}{'author'}{'name'};
 266                 my $summary = $pdata->{$module}{$id}{'summary'};
 267                 $pbugs->{$bug}{'summary'} = $summary;
 268                 $pbugs->{$bug}{'author'}{$author} = 1;
 269             }
 270         }
 271     }
 272
 273     # try to replace summaries with bug names from bugzilla
 274     if ($check_bugzilla) {
 275         print "Getting bug titles:\n";
 276         foreach my $bug ( sort { $a cmp $b } keys %{$pbugs}) {
 277             $pbugs->{$bug}{'summary'} = get_bug_name($bug, $pbugs->{$bug}{'summary'});
 278         }
 279     }
 280 }
 281
 282 sub open_log_file($$$$$$)
 283 {
 284     my ($log_dir, $log_prefix, $log_suffix, $top_dir, $branch_name, $wiki) = @_;
 285
 286     my $logfilename = "$log_prefix-$branch_name-$log_suffix";
 287     $logfilename = "$log_dir/$logfilename" if (defined $log_dir);
 288     if ($wiki) {
 289         $logfilename .= ".wiki";
 290     } else {
 291         $logfilename .= ".log";
 292     }
 293
 294     if (-f $logfilename) {
 295         print "WARNING: The log file already exists: $logfilename\n";
 296         print "Do you want to overwrite it? (Y/n)?\n";
 297         my $answer = <STDIN>;
 298         chomp $answer;
 299         $answer = "y" unless ($answer);
 300         die "Please, rename the file or choose another log suffix\n" if ( lc($answer) ne "y" );
 301     }
 302
 303     my $log;
 304     open($log, '>', $logfilename) || die "Can't open \"$logfilename\" for writing: $!\n";
 305
 306     return $log;
 307 }
 308
 309 sub print_commit_summary($$$$$$)
 310 {
 311     my ($summary, $pmodule_title, $pbugs, $pauthors, $prefix, $log) = @_;
 312
 313     return if ( $summary eq "" );
 314
 315     # print module title if not done yet
 316     if ( defined ${$pmodule_title} ) {
 317         print $log "${$pmodule_title}\n";
 318         ${$pmodule_title} = undef;
 319     }
 320
 321     # finally print the summary line
 322     my $bugs = "";
 323     if ( %{$pbugs} ) {
 324         $bugs = " (" . join (", ", keys %{$pbugs}) . ")";
 325     }
 326
 327     my $authors = "";
 328     if ( %{$pauthors} ) {
 329         $authors = " [" . join (", ", keys %{$pauthors}) . "]";
 330     }
 331
 332     print $log $prefix, $summary, $bugs, $authors, "\n";
 333 }
 334
 335 sub print_commits($$$)
 336 {
 337     my ($pdata, $log, $wiki) = @_;
 338
 339     foreach my $module ( sort { $a cmp $b } keys %{$pdata}) {
 340         # check if this module has any entries at all
 341         my $module_title = "+ $module";
 342         if ( %{$pdata->{$module}} ) {
 343             my $old_summary="";
 344             my %authors = ();
 345             my %bugs = ();
 346             foreach my $id ( sort { lc $pdata->{$module}{$a}{'summary'} cmp lc $pdata->{$module}{$b}{'summary'} } keys %{$pdata->{$module}}) {
 347                 my $summary = $pdata->{$module}{$id}{'summary'};
 348                 if ($summary ne $old_summary) {
 349                     print_commit_summary($old_summary, \$module_title, \%bugs, \%authors, "    + ", $log);
 350                     $old_summary = $summary;
 351                     %authors = ();
 352                     %bugs = ();
 353                 }
 354                 # collect bug numbers
 355                 if (defined $pdata->{$module}{$id}{'bugs'}) {
 356                     foreach my $bug (keys %{$pdata->{$module}{$id}{'bugs'}}) {
 357                         $bugs{$bug} = 1;
 358                     }
 359                 }
 360                 # collect author names
 361                 my $author = $pdata->{$module}{$id}{'author'}{'name'};
 362                 $authors{$author} = 1;
 363             }
 364             print_commit_summary($old_summary, \$module_title, \%bugs, \%authors, "    + ", $log);
 365         }
 366     }
 367 }
 368
 369 sub get_bug_name($$)
 370 {
 371     my ($bug, $summary) = @_;
 372     print "$bug: ";
 373
 374     $bug =~ m/(?:(\w*)\#+(\d+))/; # fdo#12345
 375     my $bugzilla = $1;            # fdo
 376     my $bug_number = $2;          # 12345
 377
 378     if ( $bugzillas{$bugzilla} ) {
 379         my $url = $bugzillas{$bugzilla} . $bug_number;
 380         my $ua = LWP::UserAgent->new;
 381         $ua->timeout(10);
 382         $ua->env_proxy;
 383         my $response = $ua->get($url);
 384         if ($response->is_success) {
 385             my $title = $response->title;
 386             if ( $title =~ s/^Bug \d+ \S+ // ) {
 387                 print "$title\n";
 388                 return $title;
 389             } else {
 390                 print "warning: not found; using commit message (only got $title)\n";
 391             }
 392         } else {
 393             print "\n";
 394         }
 395     } else {
 396         print "\n";
 397     }
 398
 399     return $summary;
 400 }
 401
 402 sub print_bugs($$$$)
 403 {
 404     my ($pbugs, $log, $wiki) = @_;
 405
 406     foreach my $bug ( sort { $a cmp $b } keys %{$pbugs}) {
 407         my $summary = $pbugs->{$bug}{'summary'};
 408
 409         my $authors = "";
 410         if ( %{$pbugs->{$bug}{'author'}} ) {
 411             $authors = " [" . join (", ", keys %{$pbugs->{$bug}{'author'}}) . "]";
 412         }
 413
 414         $bug =~ s/(.*)\#(.*)/* {{$1|$2}}/ if ($wiki);
 415         print $log $bug, " ", $summary, $authors, "\n";
 416     }
 417 }
 418
 419 sub print_bugs_changelog($$$$)
 420 {
 421     my ($pbugs, $log, $wiki) = @_;
 422
 423     foreach my $bug ( sort { $a cmp $b } keys %{$pbugs}) {
 424         my $summary = $pbugs->{$bug}{'summary'};
 425
 426         my $authors = "";
 427         if ( %{$pbugs->{$bug}{'author'}} ) {
 428             $authors = " [" . join (", ", keys %{$pbugs->{$bug}{'author'}}) . "]";
 429         }
 430
 431         print $log "    + $summary ($bug)$authors\n";
 432     }
 433 }
 434
 435 sub print_bugnumbers($$$$)
 436 {
 437     my ($pbugs, $log, $wiki) = @_;
 438
 439     print $log join ("\n", sort { $a cmp $b } keys %{$pbugs}), "\n";
 440 }
 441
 442 sub generate_log($$$$$$$$)
 443 {
 444     my ($pused_data, $print_func, $log_dir, $log_prefix, $log_suffix, $top_dir, $branch_name, $wiki) = @_;
 445
 446     my $log = open_log_file($log_dir, $log_prefix, $log_suffix, $top_dir, $branch_name, $wiki);
 447     & {$print_func} ($pused_data, $log, $wiki);
 448     close $log;
 449 }
 450
 451 ########################################################################
 452 # help
 453
 454 sub usage()
 455 {
 456     print "This script generates LO git commit summary\n\n" .
 457
 458           "Usage: lo-commit-stat [--help] [--no-submodules] [--module=<module>] --log-dir=<dir> --log-suffix=<string> topdir [git_arg...]\n\n" .
 459
 460           "Options:\n" .
 461           "     --help           print this help\n" .
 462           "     --no-submodule   read changes just from the main repository, ignore submodules\n" .
 463           "     --module=<module> summarize just changes from the given module, use \"core\"\n" .
 464           "                      for the main module\n" .
 465           "     --log-dir=<dir>  directory where to put the generated log\n" .
 466           "     --log-suffix=<string> suffix of the log file name; the result will be\n" .
 467           "                      commit-log-<branch>-<log-name-suffix>.log; the branch name\n" .
 468           "                      is detected automatically\n" .
 469           "     --commits        generete log with all commits (default)\n" .
 470           "     --bugs           generate log with bugzilla entries\n" .
 471           "     --bugs-changelog generate log with bugzilla entries, use changelog style\n" .
 472           "     --bugs-wiki      generate log with bugzilla entries, use wiki markup\n" .
 473           "     --bugs-numbers   generate log with bugzilla numbers\n" .
 474           "     --rev-list       use \"git rev-list\" instead of \"git log\"; useful to check\n" .
 475           "                      differences between branches\n" .
 476           "     --cherry         use \"git cherry\" instead of \"git log\"; detects cherry-picked\n" .
 477           "                      commits between branches\n" .
 478           "      topdir          directory with the libreoffice/core clone\n" .
 479           "      git_arg         extra parameters passed to the git command to define\n" .
 480           "                      the area of interest; The default command is \"git log\" and\n" .
 481           "                      parameters might be, for example, --after=\"2010-09-27\" or\n" .
 482           "                      TAG..HEAD; with the option --rev-list, useful might be, for\n" .
 483           "                      example origin/master ^origin/libreoffice-3-3; with the option\n" .
 484           "                      --rev-list, useful might be, for example libreoffice-3.6.3.2\n" .
 485           "                      libreoffice-3.6.4.1\n";
 486 }
 487
 488
 489 #######################################################################
 490 #######################################################################
 491 # MAIN
 492 #######################################################################
 493 #######################################################################
 494
 495
 496 my $module;
 497 my %generate_log = ();
 498 my $top_dir;
 499 my $log_dir;
 500 my $log_suffix;
 501 my $log;
 502 my $list_bugs = 0;
 503 my $check_bugzilla = 0;
 504 my $branch_name;
 505 my $git_command = "git log";
 506 my $git_cherry;
 507 my $git_args = "";
 508 my $branch_name;
 509 my %data;
 510 my %bugs = ();
 511
 512
 513 foreach my $arg (@ARGV) {
 514     if ($arg eq '--help') {
 515         usage();
 516         exit;
 517     } elsif ($arg eq '--no-submodule') {
 518         $module = "core";
 519     } elsif ($arg =~ m/--module=(.*)/) {
 520         $module = $1;
 521     } elsif ($arg =~ m/--log-suffix=(.*)/) {
 522         $log_suffix = "$1";
 523     } elsif ($arg =~ m/--log-dir=(.*)/) {
 524         $log_dir = "$1";
 525     } elsif ($arg eq '--commits') {
 526         $generate_log{"commits"} = 1;
 527     } elsif ($arg eq '--bugs') {
 528         $generate_log{"bugs"} = 1;
 529         $check_bugzilla = 1;
 530         $list_bugs = 1;
 531     } elsif ($arg eq '--bugs-changelog') {
 532         $generate_log{"bugs-changelog"} = 1;
 533         $check_bugzilla = 1;
 534         $list_bugs = 1;
 535     } elsif ($arg eq '--bugs-wiki' || $arg eq '--wikibugs') {
 536         $generate_log{"bugs-wiki"} = 1;
 537         $check_bugzilla = 1;
 538         $list_bugs = 1;
 539     } elsif ($arg eq '--bugs-numbers' || $arg eq '--bug-numbers') {
 540         $generate_log{"bugs-numbers"} = 1;
 541         $list_bugs = 1;
 542     } elsif ($arg eq '--rev-list') {
 543         $git_command = "git rev-list --pretty=medium"
 544     } elsif ($arg eq '--cherry') {
 545         $git_command = "git log";
 546         $git_cherry = 1;
 547     } else {
 548         if (! defined $top_dir) {
 549             $top_dir=$arg;
 550         } else {
 551             $git_args .= " $arg";
 552         }
 553     }
 554 }
 555
 556 # default log
 557 if (%generate_log == 0) {
 558     $generate_log{"commits"} = 1;
 559 }
 560
 561 # we want only one module
 562 if ($module) {
 563     my $name = $module_dirname{$module};
 564     %module_dirname = ();
 565     $module_dirname{$module} = $name;
 566 }
 567
 568 (defined $top_dir) || die "Error: top directory is not defined\n";
 569 (-d "$top_dir") || die "Error: not a directory: $top_dir\n";
 570 (-f "$top_dir/.git/config") || die "Error: can't find $top_dir/.git/config\n";
 571
 572 (!defined $log_dir) || (-d $log_dir) || die "Error: directory does no exist: $log_dir\n";
 573
 574 (defined $log_suffix) || die "Error: define log suffix using --log-suffix=<string>\n";
 575
 576 $branch_name = get_branch_name($top_dir);
 577
 578 load_data(\%data, $top_dir, \%module_dirname, $branch_name, $git_command, $git_cherry, $git_args);
 579 get_bug_list(\%data, \%bugs, $check_bugzilla) if ($list_bugs);
 580
 581 generate_log(\%data, \&print_commits,        $log_dir, "commits",        $log_suffix, $top_dir, $branch_name, 0) if (defined $generate_log{"commits"});
 582 generate_log(\%bugs, \&print_bugs,           $log_dir, "bugs",           $log_suffix, $top_dir, $branch_name, 0) if (defined $generate_log{"bugs"});
 583 generate_log(\%bugs, \&print_bugs,           $log_dir, "bugs",           $log_suffix, $top_dir, $branch_name, 1) if (defined $generate_log{"bugs-wiki"});
 584 generate_log(\%bugs, \&print_bugs_changelog, $log_dir, "bugs-changelog", $log_suffix, $top_dir, $branch_name, 0) if (defined $generate_log{"bugs-changelog"});
 585 generate_log(\%bugs, \&print_bugnumbers,     $log_dir, "bug-numbers",    $log_suffix, $top_dir, $branch_name, 0) if (defined $generate_log{"bugs-numbers"});