etc/update-copyright.py

   1 #!/usr/bin/python
   2 #
   3 # Copyright (C) 2013-2017 Free Software Foundation, Inc.
   4 #
   5 # This script is free software; you can redistribute it and/or modify
   6 # it under the terms of the GNU General Public License as published by
   7 # the Free Software Foundation; either version 3, or (at your option)
   8 # any later version.
   9
  10 # This script adjusts the copyright notices at the top of source files
  11 # so that they have the form:
  12 #
  13 #   Copyright XXXX-YYYY Free Software Foundation, Inc.
  14 #
  15 # It doesn't change code that is known to be maintained elsewhere or
  16 # that carries a non-FSF copyright.
  17 #
  18 # Pass --this-year to the script if you want it to add the current year
  19 # to all applicable notices.  Pass --quilt if you are using quilt and
  20 # want files to be added to the quilt before being changed.
  21 #
  22 # By default the script will update all directories for which the
  23 # output has been vetted.  You can instead pass the names of individual
  24 # directories, including those that haven't been approved.  So:
  25 #
  26 #    update-copyright.pl --this-year
  27 #
  28 # is the command that would be used at the beginning of a year to update
  29 # all copyright notices (and possibly at other times to check whether
  30 # new files have been added with old years).  On the other hand:
  31 #
  32 #    update-copyright.pl --this-year libjava
  33 #
  34 # would run the script on just libjava/.
  35 #
  36 # This script was copied from gcc's contrib/ and modified to suit
  37 # binutils.  In contrast to the gcc script, this one will update
  38 # the testsuite and --version output strings too.
  39
  40 import os
  41 import re
  42 import sys
  43 import time
  44 import subprocess
  45
  46 class Errors:
  47     def __init__ (self):
  48         self.num_errors = 0
  49
  50     def report (self, filename, string):
  51         if filename:
  52             string = filename + ': ' + string
  53         sys.stderr.write (string + '\n')
  54         self.num_errors += 1
  55
  56     def ok (self):
  57         return self.num_errors == 0
  58
  59 class GenericFilter:
  60     def __init__ (self):
  61         self.skip_files = set()
  62         self.skip_dirs = set()
  63         self.skip_extensions = set()
  64         self.fossilised_files = set()
  65         self.own_files = set()
  66
  67         self.skip_files |= set ([
  68                 # Skip licence files.
  69                 'COPYING',
  70                 'COPYING.LIB',
  71                 'COPYING3',
  72                 'COPYING3.LIB',
  73                 'COPYING.LIBGLOSS',
  74                 'COPYING.NEWLIB',
  75                 'LICENSE',
  76                 'fdl.texi',
  77                 'gpl_v3.texi',
  78                 'fdl-1.3.xml',
  79                 'gpl-3.0.xml',
  80
  81                 # Skip auto- and libtool-related files
  82                 'aclocal.m4',
  83                 'compile',
  84                 'config.guess',
  85                 'config.sub',
  86                 'depcomp',
  87                 'install-sh',
  88                 'libtool.m4',
  89                 'ltmain.sh',
  90                 'ltoptions.m4',
  91                 'ltsugar.m4',
  92                 'ltversion.m4',
  93                 'lt~obsolete.m4',
  94                 'missing',
  95                 'mkdep',
  96                 'mkinstalldirs',
  97                 'move-if-change',
  98                 'shlibpath.m4',
  99                 'symlink-tree',
 100                 'ylwrap',
 101
 102                 # Skip FSF mission statement, etc.
 103                 'gnu.texi',
 104                 'funding.texi',
 105                 'appendix_free.xml',
 106
 107                 # Skip imported texinfo files.
 108                 'texinfo.tex',
 109                 ])
 110
 111         self.skip_extensions |= set ([
 112                 # Maintained by the translation project.
 113                 '.po',
 114
 115                 # Automatically-generated.
 116                 '.pot',
 117                 ])
 118
 119         self.skip_dirs |= set ([
 120                 'autom4te.cache',
 121                 ])
 122
 123
 124     def get_line_filter (self, dir, filename):
 125         if filename.startswith ('ChangeLog'):
 126             # Ignore references to copyright in changelog entries.
 127             return re.compile ('\t')
 128
 129         return None
 130
 131     def skip_file (self, dir, filename):
 132         if filename in self.skip_files:
 133             return True
 134
 135         (base, extension) = os.path.splitext (os.path.join (dir, filename))
 136         if extension in self.skip_extensions:
 137             return True
 138
 139         if extension == '.in':
 140             # Skip .in files produced by automake.
 141             if os.path.exists (base + '.am'):
 142                 return True
 143
 144             # Skip files produced by autogen
 145             if (os.path.exists (base + '.def')
 146                 and os.path.exists (base + '.tpl')):
 147                 return True
 148
 149         # Skip configure files produced by autoconf
 150         if filename == 'configure':
 151             if os.path.exists (base + '.ac'):
 152                 return True
 153             if os.path.exists (base + '.in'):
 154                 return True
 155
 156         return False
 157
 158     def skip_dir (self, dir, subdir):
 159         return subdir in self.skip_dirs
 160
 161     def is_fossilised_file (self, dir, filename):
 162         if filename in self.fossilised_files:
 163             return True
 164         # Only touch current current ChangeLogs.
 165         if filename != 'ChangeLog' and filename.find ('ChangeLog') >= 0:
 166             return True
 167         return False
 168
 169     def by_package_author (self, dir, filename):
 170         return filename in self.own_files
 171
 172 class Copyright:
 173     def __init__ (self, errors):
 174         self.errors = errors
 175
 176         # Characters in a range of years.  Include '.' for typos.
 177         ranges = '[0-9](?:[-0-9.,\s]|\s+and\s+)*[0-9]'
 178
 179         # Non-whitespace characters in a copyright holder's name.
 180         name = '[\w.,-]'
 181
 182         # Matches one year.
 183         self.year_re = re.compile ('[0-9]+')
 184
 185         # Matches part of a year or copyright holder.
 186         self.continuation_re = re.compile (ranges + '|' + name)
 187
 188         # Matches a full copyright notice:
 189         self.copyright_re = re.compile (
 190             # 1: 'Copyright (C)', etc.
 191             '([Cc]opyright'
 192             '|[Cc]opyright\s+\([Cc]\)'
 193             '|[Cc]opyright\s+%s'
 194             '|[Cc]opyright\s+&copy;'
 195             '|[Cc]opyright\s+@copyright{}'
 196             '|@set\s+copyright[\w-]+)'
 197
 198             # 2: the years.  Include the whitespace in the year, so that
 199             # we can remove any excess.
 200             '(\s*(?:' + ranges + ',?'
 201             '|@value\{[^{}]*\})\s*)'
 202
 203             # 3: 'by ', if used
 204             '(by\s+)?'
 205
 206             # 4: the copyright holder.  Don't allow multiple consecutive
 207             # spaces, so that right-margin gloss doesn't get caught
 208             # (e.g. gnat_ugn.texi).
 209             '(' + name + '(?:\s?' + name + ')*)?')
 210
 211         # A regexp for notices that might have slipped by.  Just matching
 212         # 'copyright' is too noisy, and 'copyright.*[0-9]' falls foul of
 213         # HTML header markers, so check for 'copyright' and two digits.
 214         self.other_copyright_re = re.compile ('(^|[^\._])copyright[^=]*[0-9][0-9]',
 215                                               re.IGNORECASE)
 216         self.comment_re = re.compile('#+|[*]+|;+|%+|//+|@c |dnl ')
 217         self.holders = { '@copying': '@copying' }
 218         self.holder_prefixes = set()
 219
 220         # True to 'quilt add' files before changing them.
 221         self.use_quilt = False
 222
 223         # If set, force all notices to include this year.
 224         self.max_year = None
 225
 226         # Goes after the year(s).  Could be ', '.
 227         self.separator = ' '
 228
 229     def add_package_author (self, holder, canon_form = None):
 230         if not canon_form:
 231             canon_form = holder
 232         self.holders[holder] = canon_form
 233         index = holder.find (' ')
 234         while index >= 0:
 235             self.holder_prefixes.add (holder[:index])
 236             index = holder.find (' ', index + 1)
 237
 238     def add_external_author (self, holder):
 239         self.holders[holder] = None
 240
 241     class BadYear():
 242         def __init__ (self, year):
 243             self.year = year
 244
 245         def __str__ (self):
 246             return 'unrecognised year: ' + self.year
 247
 248     def parse_year (self, string):
 249         year = int (string)
 250         if len (string) == 2:
 251             if year > 70:
 252                 return year + 1900
 253         elif len (string) == 4:
 254             return year
 255         raise self.BadYear (string)
 256
 257     def year_range (self, years):
 258         year_list = [self.parse_year (year)
 259                      for year in self.year_re.findall (years)]
 260         assert len (year_list) > 0
 261         return (min (year_list), max (year_list))
 262
 263     def set_use_quilt (self, use_quilt):
 264         self.use_quilt = use_quilt
 265
 266     def include_year (self, year):
 267         assert not self.max_year
 268         self.max_year = year
 269
 270     def canonicalise_years (self, dir, filename, filter, years):
 271         # Leave texinfo variables alone.
 272         if years.startswith ('@value'):
 273             return years
 274
 275         (min_year, max_year) = self.year_range (years)
 276
 277         # Update the upper bound, if enabled.
 278         if self.max_year and not filter.is_fossilised_file (dir, filename):
 279             max_year = max (max_year, self.max_year)
 280
 281         # Use a range.
 282         if min_year == max_year:
 283             return '%d' % min_year
 284         else:
 285             return '%d-%d' % (min_year, max_year)
 286
 287     def strip_continuation (self, line):
 288         line = line.lstrip()
 289         match = self.comment_re.match (line)
 290         if match:
 291             line = line[match.end():].lstrip()
 292         return line
 293
 294     def is_complete (self, match):
 295         holder = match.group (4)
 296         return (holder
 297                 and (holder not in self.holder_prefixes
 298                      or holder in self.holders))
 299
 300     def update_copyright (self, dir, filename, filter, file, line, match):
 301         orig_line = line
 302         next_line = None
 303         pathname = os.path.join (dir, filename)
 304
 305         intro = match.group (1)
 306         if intro.startswith ('@set'):
 307             # Texinfo year variables should always be on one line
 308             after_years = line[match.end (2):].strip()
 309             if after_years != '':
 310                 self.errors.report (pathname,
 311                                     'trailing characters in @set: '
 312                                     + after_years)
 313                 return (False, orig_line, next_line)
 314         else:
 315             # If it looks like the copyright is incomplete, add the next line.
 316             while not self.is_complete (match):
 317                 try:
 318                     next_line = file.next()
 319                 except StopIteration:
 320                     break
 321
 322                 # If the next line doesn't look like a proper continuation,
 323                 # assume that what we've got is complete.
 324                 continuation = self.strip_continuation (next_line)
 325                 if not self.continuation_re.match (continuation):
 326                     break
 327
 328                 # Merge the lines for matching purposes.
 329                 orig_line += next_line
 330                 line = line.rstrip() + ' ' + continuation
 331                 next_line = None
 332
 333                 # Rematch with the longer line, at the original position.
 334                 match = self.copyright_re.match (line, match.start())
 335                 assert match
 336
 337             holder = match.group (4)
 338
 339             # Use the filter to test cases where markup is getting in the way.
 340             if filter.by_package_author (dir, filename):
 341                 assert holder not in self.holders
 342
 343             elif not holder:
 344                 self.errors.report (pathname, 'missing copyright holder')
 345                 return (False, orig_line, next_line)
 346
 347             elif holder not in self.holders:
 348                 self.errors.report (pathname,
 349                                     'unrecognised copyright holder: ' + holder)
 350                 return (False, orig_line, next_line)
 351
 352             else:
 353                 # See whether the copyright is associated with the package
 354                 # author.
 355                 canon_form = self.holders[holder]
 356                 if not canon_form:
 357                     return (False, orig_line, next_line)
 358
 359                 # Make sure the author is given in a consistent way.
 360                 line = (line[:match.start (4)]
 361                         + canon_form
 362                         + line[match.end (4):])
 363
 364                 # Remove any 'by'
 365                 line = line[:match.start (3)] + line[match.end (3):]
 366
 367         # Update the copyright years.
 368         years = match.group (2).strip()
 369         if (self.max_year
 370             and match.start(0) > 0 and line[match.start(0)-1] == '"'
 371             and not filter.is_fossilised_file (dir, filename)):
 372             # A printed copyright date consists of the current year
 373             canon_form = '%d' % self.max_year
 374         else:
 375             try:
 376                 canon_form = self.canonicalise_years (dir, filename, filter, years)
 377             except self.BadYear as e:
 378                 self.errors.report (pathname, str (e))
 379                 return (False, orig_line, next_line)
 380
 381         line = (line[:match.start (2)]
 382                 + ' ' + canon_form + self.separator
 383                 + line[match.end (2):])
 384
 385         # Use the standard (C) form.
 386         if intro.endswith ('right'):
 387             intro += ' (C)'
 388         elif intro.endswith ('(c)'):
 389             intro = intro[:-3] + '(C)'
 390         line = line[:match.start (1)] + intro + line[match.end (1):]
 391
 392         # Strip trailing whitespace
 393         line = line.rstrip() + '\n'
 394
 395         return (line != orig_line, line, next_line)
 396
 397     def process_file (self, dir, filename, filter):
 398         pathname = os.path.join (dir, filename)
 399         if filename.endswith ('.tmp'):
 400             # Looks like something we tried to create before.
 401             try:
 402                 os.remove (pathname)
 403             except OSError:
 404                 pass
 405             return
 406
 407         lines = []
 408         changed = False
 409         line_filter = filter.get_line_filter (dir, filename)
 410         with open (pathname, 'r') as file:
 411             prev = None
 412             for line in file:
 413                 while line:
 414                     next_line = None
 415                     # Leave filtered-out lines alone.
 416                     if not (line_filter and line_filter.match (line)):
 417                         match = self.copyright_re.search (line)
 418                         if match:
 419                             res = self.update_copyright (dir, filename, filter,
 420                                                          file, line, match)
 421                             (this_changed, line, next_line) = res
 422                             changed = changed or this_changed
 423
 424                         # Check for copyright lines that might have slipped by.
 425                         elif self.other_copyright_re.search (line):
 426                             self.errors.report (pathname,
 427                                                 'unrecognised copyright: %s'
 428                                                 % line.strip())
 429                     lines.append (line)
 430                     line = next_line
 431
 432         # If something changed, write the new file out.
 433         if changed and self.errors.ok():
 434             tmp_pathname = pathname + '.tmp'
 435             with open (tmp_pathname, 'w') as file:
 436                 for line in lines:
 437                     file.write (line)
 438             if self.use_quilt:
 439                 subprocess.call (['quilt', 'add', pathname])
 440             os.rename (tmp_pathname, pathname)
 441
 442     def process_tree (self, tree, filter):
 443         for (dir, subdirs, filenames) in os.walk (tree):
 444             # Don't recurse through directories that should be skipped.
 445             for i in xrange (len (subdirs) - 1, -1, -1):
 446                 if filter.skip_dir (dir, subdirs[i]):
 447                     del subdirs[i]
 448
 449             # Handle the files in this directory.
 450             for filename in filenames:
 451                 if filter.skip_file (dir, filename):
 452                     sys.stdout.write ('Skipping %s\n'
 453                                       % os.path.join (dir, filename))
 454                 else:
 455                     self.process_file (dir, filename, filter)
 456
 457 class CmdLine:
 458     def __init__ (self, copyright = Copyright):
 459         self.errors = Errors()
 460         self.copyright = copyright (self.errors)
 461         self.dirs = []
 462         self.default_dirs = []
 463         self.chosen_dirs = []
 464         self.option_handlers = dict()
 465         self.option_help = []
 466
 467         self.add_option ('--help', 'Print this help', self.o_help)
 468         self.add_option ('--quilt', '"quilt add" files before changing them',
 469                          self.o_quilt)
 470         self.add_option ('--this-year', 'Add the current year to every notice',
 471                          self.o_this_year)
 472
 473     def add_option (self, name, help, handler):
 474         self.option_help.append ((name, help))
 475         self.option_handlers[name] = handler
 476
 477     def add_dir (self, dir, filter = GenericFilter()):
 478         self.dirs.append ((dir, filter))
 479
 480     def o_help (self, option = None):
 481         sys.stdout.write ('Usage: %s [options] dir1 dir2...\n\n'
 482                           'Options:\n' % sys.argv[0])
 483         format = '%-15s %s\n'
 484         for (what, help) in self.option_help:
 485             sys.stdout.write (format % (what, help))
 486         sys.stdout.write ('\nDirectories:\n')
 487
 488         format = '%-25s'
 489         i = 0
 490         for (dir, filter) in self.dirs:
 491             i += 1
 492             if i % 3 == 0 or i == len (self.dirs):
 493                 sys.stdout.write (dir + '\n')
 494             else:
 495                 sys.stdout.write (format % dir)
 496         sys.exit (0)
 497
 498     def o_quilt (self, option):
 499         self.copyright.set_use_quilt (True)
 500
 501     def o_this_year (self, option):
 502         self.copyright.include_year (time.localtime().tm_year)
 503
 504     def main (self):
 505         for arg in sys.argv[1:]:
 506             if arg[:1] != '-':
 507                 self.chosen_dirs.append (arg)
 508             elif arg in self.option_handlers:
 509                 self.option_handlers[arg] (arg)
 510             else:
 511                 self.errors.report (None, 'unrecognised option: ' + arg)
 512         if self.errors.ok():
 513             if len (self.chosen_dirs) == 0:
 514                 self.chosen_dirs = self.default_dirs
 515             if len (self.chosen_dirs) == 0:
 516                 self.o_help()
 517             else:
 518                 for chosen_dir in self.chosen_dirs:
 519                     canon_dir = os.path.join (chosen_dir, '')
 520                     count = 0
 521                     for (dir, filter) in self.dirs:
 522                         if (dir + os.sep).startswith (canon_dir):
 523                             count += 1
 524                             self.copyright.process_tree (dir, filter)
 525                     if count == 0:
 526                         self.errors.report (None, 'unrecognised directory: '
 527                                             + chosen_dir)
 528         sys.exit (0 if self.errors.ok() else 1)
 529
 530 #----------------------------------------------------------------------------
 531
 532 class TopLevelFilter (GenericFilter):
 533     def skip_dir (self, dir, subdir):
 534         return True
 535
 536 class ConfigFilter (GenericFilter):
 537     def __init__ (self):
 538         GenericFilter.__init__ (self)
 539
 540     def skip_file (self, dir, filename):
 541         if filename.endswith ('.m4'):
 542             pathname = os.path.join (dir, filename)
 543             with open (pathname) as file:
 544                 # Skip files imported from gettext.
 545                 if file.readline().find ('gettext-') >= 0:
 546                     return True
 547         return GenericFilter.skip_file (self, dir, filename)
 548
 549 class LdFilter (GenericFilter):
 550     def __init__ (self):
 551         GenericFilter.__init__ (self)
 552
 553         self.skip_extensions |= set ([
 554                 # ld testsuite output match files.
 555                 '.ro',
 556                 ])
 557
 558 class BinutilsCopyright (Copyright):
 559     def __init__ (self, errors):
 560         Copyright.__init__ (self, errors)
 561
 562         canon_fsf = 'Free Software Foundation, Inc.'
 563         self.add_package_author ('Free Software Foundation', canon_fsf)
 564         self.add_package_author ('Free Software Foundation.', canon_fsf)
 565         self.add_package_author ('Free Software Foundation Inc.', canon_fsf)
 566         self.add_package_author ('Free Software Foundation, Inc', canon_fsf)
 567         self.add_package_author ('Free Software Foundation, Inc.', canon_fsf)
 568         self.add_package_author ('The Free Software Foundation', canon_fsf)
 569         self.add_package_author ('The Free Software Foundation, Inc.', canon_fsf)
 570         self.add_package_author ('Software Foundation, Inc.', canon_fsf)
 571
 572         self.add_external_author ('Carnegie Mellon University')
 573         self.add_external_author ('John D. Polstra.')
 574         self.add_external_author ('Linaro Ltd.')
 575         self.add_external_author ('MIPS Computer Systems, Inc.')
 576         self.add_external_author ('Red Hat Inc.')
 577         self.add_external_author ('Regents of the University of California.')
 578         self.add_external_author ('The Regents of the University of California.')
 579         self.add_external_author ('Third Eye Software, Inc.')
 580         self.add_external_author ('Ulrich Drepper')
 581         self.add_external_author ('Synopsys Inc.')
 582
 583 class BinutilsCmdLine (CmdLine):
 584     def __init__ (self):
 585         CmdLine.__init__ (self, BinutilsCopyright)
 586
 587         self.add_dir ('.', TopLevelFilter())
 588         self.add_dir ('bfd')
 589         self.add_dir ('binutils')
 590         self.add_dir ('config', ConfigFilter())
 591         self.add_dir ('cpu')
 592         self.add_dir ('elfcpp')
 593         self.add_dir ('etc')
 594         self.add_dir ('gas')
 595         self.add_dir ('gdb')
 596         self.add_dir ('gold')
 597         self.add_dir ('gprof')
 598         self.add_dir ('include')
 599         self.add_dir ('ld', LdFilter())
 600         self.add_dir ('libdecnumber')
 601         self.add_dir ('libiberty')
 602         self.add_dir ('opcodes')
 603         self.add_dir ('readline')
 604         self.add_dir ('sim')
 605
 606         self.default_dirs = [
 607             'bfd',
 608             'binutils',
 609             'elfcpp',
 610             'etc',
 611             'gas',
 612             'gold',
 613             'gprof',
 614             'include',
 615             'ld',
 616             'libiberty',
 617             'opcodes',
 618             ]
 619
 620 BinutilsCmdLine().main()