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";
28 usage() unless(@ARGV);
29 usage() unless GetOptions("scan|s" => \$scan,
31 "write|w" => \$writecopyright,
32 "help|h" => sub { usage(); });
37 write_copyright() if ($merge || $writecopyright);
42 for my $file (@filenames)
60 die("$me: no files known, run $0 --scan\n");
62 unless(copy($COPYRIGHTSTUB, "debian/copyright"))
64 die("$me: cannot copy $COPYRIGHTSTUB to debian/copyright: $!\n");
66 unless(open(COPYRIGHT, ">>debian/copyright"))
68 die("$me: cannot append to debian/copyright: $!\n");
71 # group files by license/license_text/copyright
73 for my $file (sort keys(%$files))
75 my $license=$files->{$file}->{license};
76 my $copyright=$files->{$file}->{copyright};
77 my $license_text=$files->{$file}->{license_text};
78 push(@{$licenses->{$license}->{$license_text}->{$copyright}}, $file);
82 for my $license (sort keys(%$licenses))
84 for my $license_text (sort keys(%{$licenses->{$license}}))
86 my $licensestr=$license;
87 if(length($license_text))
90 $licensestr .= " (REF$refnum)";
91 $refs{$licensestr}=$license_text;
93 for my $copyright (sort keys(%{$licenses->{$license}->{$license_text}}))
95 next if(!length($license) && !length($copyright) && !length($license_text));
96 my @filelist=sort @{$licenses->{$license}->{$license_text}->{$copyright}};
97 print COPYRIGHT "Files: ", join(', ', @filelist), "\n";
98 print COPYRIGHT "Copyright: $copyright\n" if length($copyright);
99 print COPYRIGHT "License: $licensestr\n" if length($licensestr);
100 print COPYRIGHT "\n";
104 for my $ref (keys(%refs))
106 print COPYRIGHT "License: $ref\n";
107 my @text=split(/\n/, $refs{$ref});
108 print COPYRIGHT map { " " . $_ . "\n" } @text;
109 print COPYRIGHT "\n";
111 print COPYRIGHT license_trailer(sort keys(%$licenses));
117 my @licenses_used=@_;
119 "Apache-2.0" => "Apache License Version 2.0",
120 "Artistic" => "Artistic License",
121 "GFDL-1.2" => "GNU Free Documentation License Version 1.2",
122 "GFDL-1.3" => "GNU Free Documentation License Version 1.3",
123 "GFDL" => "GNU Free Documentation License",
124 "GPL-2" => "GNU General Public License Version 2",
125 "GPL-3" => "GNU General Public License Version 3",
126 "GPL" => "GNU General Public License",
127 "LGPL-2" => "GNU Library General Public License Version 2",
128 "LGPL-2.1" => "GNU Lesser General Public License Version 2.1",
129 "LGPL-3" => "GNU Lesser General Public License Version 3",
130 "LGPL" => "GNU Lesser General Public License",
134 TYPE: for my $type (keys(%$license_data))
136 for my $license (@licenses_used)
138 if($license =~ /$type(\+|\b)/i)
140 $types_found{$type}=1;
141 # avoid matching eg GPL-2 *and* GPL
147 for my $type (sort keys(%types_found))
149 $text .= "The complete text of the " . $license_data->{$type} ." can be\n";
150 $text .= "found in /usr/share/common-licenses/$type\n";
161 if(open(LICENSECHECK, "licensecheck --copyright \"$file\"|"))
163 while(<LICENSECHECK>)
176 $copyright =~ s/^\s*Copyright\s*:\s*//;
177 $license =~ s/.*UNKNOWN.*//;
178 $license =~ s/(L?GPL) \(v([\.\d]+) or later\)/$1-$2+/i;
179 $license =~ s/(L?GPL) \(v([\.\d]+)\)/$1-$2/i;
180 $license =~ s/G?FDL \(v([\.\d]+) or later\)/GFDL-$1+/i;
181 $license =~ s/G?FDL \(v([\.\d]+)\)/GFDL-$1/i;
182 $license =~ s/Apache \(v([\.\d]+) or later\)/Apache-$1+/i;
183 $license =~ s/Apache \(v([\.\d]+)\)/Apache-$1+/i;
185 return($license, $copyright);
191 unless(open(FILE, $filename))
193 warn("$me: $filename: cannot open: $!\n");
197 for(my $i=0; $i < 15; $i++)
204 my $hash=sha256_hex($header);
205 if( (!exists($files->{$filename})) ||
206 ($files->{$filename}->{hash} ne $hash))
208 filechanged($filename, $hash, $header);
215 my($filename, $hash, $header)=@_;
216 my($license_guess, $copyright_guess)=guess_license($filename);
219 license=>$license_guess,
220 copyright=>$copyright_guess,
223 if(exists($files->{$filename}))
225 if(exists($files->{$filename}->{copyright}))
227 $new->{$filename}->{copyright_old}=$files->{$filename}->{copyright};
229 if(exists($files->{$filename}->{license}))
231 $new->{$filename}->{license_old}=$files->{$filename}->{license};
233 if(exists($files->{$filename}->{license_text}))
235 $new->{$filename}->{license_text_old}=$files->{$filename}->{license_text};
242 if(/^\.git/ || /^\.cvs/ || /^debian/ || /^modules$/)
244 $File::Find::prune=1;
248 push(@filenames, $File::Find::name);
259 unless(open(YAML,$FILESCACHE))
261 warn("$me: cannot load cache $FILESCACHE: $!\n");
276 unless(open(YAML,">$FILESCACHE"))
278 warn("$me: cannot save cache $FILESCACHE: $!\n");
281 print YAML Dump($files);
290 warn("$me: no new/changed files found\n");
293 unless(open(NEW,">$NEWFILES"))
295 die("$me: cannot write to $NEWFILES: $!\n");
297 for my $file (sort keys %$new)
299 print NEW "File: $file\n";
300 print NEW "Hash: ", $new->{$file}->{hash}, "\n";
301 print NEW "Copyright: ", $new->{$file}->{copyright}, "\n";
302 print NEW "License: ", $new->{$file}->{license}, "\n";
303 print NEW "Licence_Text: \n";
304 if($new->{$file}->{license_old})
306 print NEW "#License_old: ", $new->{$file}->{license_old}, "\n";
308 if($new->{$file}->{copyright_old})
310 print NEW "#Copyright_old: ", $new->{$file}->{copyright_old}, "\n";
312 if($new->{$file}->{licence_text_old})
314 print NEW "#License_text_old: ", $new->{$file}->{licence_text_old}, "\n";
316 print NEW "#Header: \n";
317 my @headerlines=split(/\n/, $new->{$file}->{header});
318 @headerlines=map { "#" . $_ } grep { defined $_; } @headerlines;
319 print NEW join("\n", @headerlines);
327 unless(open(NEW, $NEWFILES))
329 die("$me: $NEWFILES: cannot open: $!\n");
336 my $in_license_text=0;
344 if($in_license_text && /^\s+(.*)/)
346 $license_text .= "\n" . $1;
348 elsif(/^File:\s*(.*)/)
351 # save previous entry
352 if(length($file) && length($hash))
354 $files->{$file}={ hash=>$hash,
355 copyright=>$copyright,
357 license_text=>$license_text };
365 elsif(/^Hash:\s*(.*)/)
369 elsif(/^Copyright:\s*(.*)/)
373 elsif(/^License:\s*(.*)/)
377 elsif(/^License_text:\s*(.*)/)
382 elsif($in_license_text && /^\s+(.*)/)
384 $license_text .= "\n" . $1;
388 warn("$me: $NEWFILES: line $line not recognized\n");
393 if(length($file) && length($hash))
395 $files->{$file}={ hash=>$hash,
396 copyright=>$copyright,
398 license_text=>$license_text };
405 my %newnames = map { $_ => 1 } @filenames;
406 for my $file (sort keys(%$files))
408 unless(exists($newnames{$file}))
410 push(@deleted, $file);
415 print "Removed files:\n";
416 print join("\n", @deleted),"\n";
433 die("usage: $me [--scan] [--merge]\n",
434 "scans for changed copyright/licenses\n",
435 " -s|-scan Scan for new files & files with changed copyright headers\n",
436 " Writes to debian/clscan/new.txt for manual review.\n",
437 " -m|--merge Merges new data from debian/clscan/new.txt\n",
438 " -w|--write Writes updated debian/copyright.\n",
439 " --merge implies --write.\n");