2 # Copyright 2009 Ian Beckwith <ianb@erislabs.net>
3 # License: GPL v2 or later.
7 $me=($0=~/(?:.*\/)?(.*)/)[0];
11 use Digest::SHA qw(sha256_hex);
15 our $CLSCANDIR="debian/clscan";
16 our $FILESCACHE="$CLSCANDIR/files.yaml";
17 our $NEWFILES="$CLSCANDIR/new.txt";
18 our $COPYRIGHTSTUB="$CLSCANDIR/copyright.in";
29 usage() unless(@ARGV);
30 usage() unless GetOptions("scan|s" => \$scan,
32 "write|w" => \$writecopyright,
33 "help|h" => sub { usage(); });
38 write_copyright() if ($merge || $writecopyright);
43 for my $file (@filenames)
61 die("$me: no files known, run $0 --scan\n");
63 unless(copy($COPYRIGHTSTUB, "debian/copyright"))
65 die("$me: cannot copy $COPYRIGHTSTUB to debian/copyright: $!\n");
67 unless(open(COPYRIGHT, ">>debian/copyright"))
69 die("$me: cannot append to debian/copyright: $!\n");
72 # group files by license/license_text/copyright
74 for my $file (sort keys(%$files))
76 my $license=$files->{$file}->{license};
77 my $copyright=$files->{$file}->{copyright};
78 my $license_text=$files->{$file}->{license_text};
79 push(@{$licenses->{$license}->{$license_text}->{$copyright}}, $file);
83 for my $license (sort keys(%$licenses))
85 for my $license_text (sort keys(%{$licenses->{$license}}))
87 my $licensestr=$license;
88 if(length($license_text))
91 # license_text + empty license = License: other
94 $licensestr = "other";
96 $licensestr .= " [REF$refnum]";
97 $refs{$licensestr}=$license_text;
99 for my $copyright (sort keys(%{$licenses->{$license}->{$license_text}}))
101 next if(!length($license) && !length($copyright) && !length($license_text));
102 my @filelist=sort @{$licenses->{$license}->{$license_text}->{$copyright}};
103 print COPYRIGHT "Files: ", join(', ', @filelist), "\n";
104 print COPYRIGHT "Copyright: $copyright\n" if length($copyright);
105 print COPYRIGHT "License: $licensestr\n" if length($licensestr);
106 print COPYRIGHT "\n";
110 for my $ref (sort keys(%refs))
112 print COPYRIGHT "License: $ref\n";
113 my @text=split(/\n/, $refs{$ref});
114 print COPYRIGHT map { " " . $_ . "\n" } @text;
115 print COPYRIGHT "\n";
117 print COPYRIGHT license_trailer(sort keys(%$licenses));
123 my @licenses_used=@_;
125 "Apache-2.0" => "Apache License Version 2.0",
126 "Artistic" => "Artistic License",
127 "GFDL-1.2" => "GNU Free Documentation License Version 1.2",
128 "GFDL-1.3" => "GNU Free Documentation License Version 1.3",
129 "GFDL" => "GNU Free Documentation License",
130 "GPL-2" => "GNU General Public License Version 2",
131 "GPL-3" => "GNU General Public License Version 3",
132 "GPL" => "GNU General Public License",
133 "LGPL-2" => "GNU Library General Public License Version 2",
134 "LGPL-2.1" => "GNU Lesser General Public License Version 2.1",
135 "LGPL-3" => "GNU Lesser General Public License Version 3",
136 "LGPL" => "GNU Lesser General Public License",
140 TYPE: for my $type (reverse sort keys(%$license_data))
142 for my $license (@licenses_used)
144 if($license =~ /$type(\+|\b)/i)
146 $types_found{$type}=1;
147 # avoid matching eg GPL-2 *and* GPL
153 # if just one, use standard style
154 if(keys(%types_found) == 1)
156 my ($file, $name)=each(%types_found);
157 $text .= "The complete text of the $name can be\n";
158 $text .= "found in /usr/share/common-licenses/$file\n";
162 # more than one, use table.
163 $text .= "The complete text of standard licenses referenced above\n";
164 $text .= "can be found in /usr/share/common-licenses/ as follows:\n\n";
165 $text .= sprintf("%-70s %s\n", "LICENSE", "FILE");
166 for my $type (sort keys(%types_found))
168 $text .= sprintf("%-70s %s\n", $license_data->{$type}, $type);
180 if(open(LICENSECHECK, "licensecheck --copyright \"$file\"|"))
182 while(<LICENSECHECK>)
195 $copyright =~ s/^\s*Copyright\s*:\s*//;
196 $license =~ s/.*UNKNOWN.*//;
197 $license =~ s/(L?GPL) \(v([\.\d]+) or later\)/$1-$2+/i;
198 $license =~ s/(L?GPL) \(v([\.\d]+)\)/$1-$2/i;
199 $license =~ s/G?FDL \(v([\.\d]+) or later\)/GFDL-$1+/i;
200 $license =~ s/G?FDL \(v([\.\d]+)\)/GFDL-$1/i;
201 $license =~ s/Apache \(v([\.\d]+) or later\)/Apache-$1+/i;
202 $license =~ s/Apache \(v([\.\d]+)\)/Apache-$1+/i;
204 return($license, $copyright);
210 unless(open(FILE, $filename))
212 warn("$me: $filename: cannot open: $!\n");
216 for(my $i=0; $i < 15; $i++)
223 my $hash=sha256_hex($header);
224 if( (!exists($files->{$filename})) ||
225 ($files->{$filename}->{hash} ne $hash))
227 filechanged($filename, $hash, $header);
234 my($filename, $hash, $header)=@_;
235 my($license_guess, $copyright_guess)=guess_license($filename);
238 license=>$license_guess,
239 copyright=>$copyright_guess,
242 if(exists($files->{$filename}))
244 if(exists($files->{$filename}->{copyright}))
246 $new->{$filename}->{copyright_old}=$files->{$filename}->{copyright};
248 if(exists($files->{$filename}->{license}))
250 $new->{$filename}->{license_old}=$files->{$filename}->{license};
252 if(exists($files->{$filename}->{license_text}))
254 $new->{$filename}->{license_text_old}=$files->{$filename}->{license_text};
261 find(\&wanted_files, ".");
266 if(/^\.git/ || /^\.cvs/ || /^debian/ || /^modules$/)
268 $File::Find::prune=1;
272 push(@filenames, $File::Find::name);
278 if(/^\./ || /^README$/ || /^COPYING$/)
280 $File::Find::prune=1;
284 push(@modules, $File::Find::name);
290 find(\&wanted_modules, "modules/");
292 for my $module (@modules)
294 unless(open(MOD, $module))
296 warn("$me: cannot open $module: $!\n");
306 $infiles = $inlicense = 0;
310 $modules{$module}->{license}=$_;
315 push(@{$modules{$module}->{files}}, $_);
333 unless(open(YAML,$FILESCACHE))
335 warn("$me: cannot load cache $FILESCACHE: $!\n");
350 unless(open(YAML,">$FILESCACHE"))
352 warn("$me: cannot save cache $FILESCACHE: $!\n");
355 print YAML Dump($files);
364 warn("$me: no new/changed files found\n");
367 unless(open(NEW,">$NEWFILES"))
369 die("$me: cannot write to $NEWFILES: $!\n");
371 for my $file (sort keys %$new)
373 print NEW "File: $file\n";
374 print NEW "Hash: ", $new->{$file}->{hash}, "\n";
375 print NEW "Copyright: ", $new->{$file}->{copyright}, "\n";
376 print NEW "License: ", $new->{$file}->{license}, "\n";
377 print NEW "License_Text: \n";
378 if($new->{$file}->{license_old})
380 print NEW "#License_old: ", $new->{$file}->{license_old}, "\n";
382 if($new->{$file}->{copyright_old})
384 print NEW "#Copyright_old: ", $new->{$file}->{copyright_old}, "\n";
386 if($new->{$file}->{licence_text_old})
388 print NEW "#License_text_old: ", $new->{$file}->{licence_text_old}, "\n";
390 print NEW "#Header: \n";
391 my @headerlines=split(/\n/, $new->{$file}->{header});
392 @headerlines=map { "#" . $_ } grep { defined $_; } @headerlines;
393 print NEW join("\n", @headerlines);
401 unless(open(NEW, $NEWFILES))
403 die("$me: $NEWFILES: cannot open: $!\n");
410 my $in_license_text=0;
417 if($in_license_text && /^\s+(.*)/)
419 $license_text .= $1 . "\n";
425 elsif(/^File:\s*(.*)/)
428 # save previous entry
429 if(length($file) && length($hash))
431 $files->{$file}={ hash=>$hash,
432 copyright=>$copyright,
434 license_text=>$license_text };
441 $in_license_text = 0;
443 elsif(/^Hash:\s*(.*)/)
447 elsif(/^Copyright:\s*(.*)/)
451 elsif(/^License:\s*(.*)/)
459 warn("$me: $NEWFILES: line $line not recognized\n");
464 if(length($file) && length($hash))
466 $files->{$file}={ hash=>$hash,
467 copyright=>$copyright,
469 license_text=>$license_text };
476 my %newnames = map { $_ => 1 } @filenames;
477 for my $file (sort keys(%$files))
479 unless(exists($newnames{$file}))
481 push(@deleted, $file);
486 print "Removed files:\n";
487 print join("\n", @deleted),"\n";
504 die("usage: $me [--scan] [--merge]\n",
505 "scans for changed copyright/licenses\n",
506 " -s|-scan Scan for new files & files with changed copyright headers\n",
507 " Writes to debian/clscan/new.txt for manual review.\n",
508 " -m|--merge Merges new data from debian/clscan/new.txt\n",
509 " -w|--write Writes updated debian/copyright.\n",
510 " --merge implies --write.\n");