git.oblomov.eu Git - rcs-fast-export/blob - rcs-fast-export.rb

   1 #!/usr/bin/ruby
   2
   3 require 'pp'
   4
   5 # Integer#odd? was introduced in Ruby 1.8.7, backport it to
   6 # older versions
   7 unless 2.respond_to? :odd?
   8         class Integer
   9                 def odd?
  10                         self % 2 == 1
  11                 end
  12         end
  13 end
  14
  15 def usage
  16         STDERR.puts <<EOM
  17 #{$0} [options] file [file ...]
  18
  19 Fast-export the RCS history of one or more file.
  20
  21 Options:
  22         --help, -h, -?          display this help text
  23         --authors-file, -A      specify a file containing username = Full Name <email> mappings
  24         --[no-]tag-each-rev     [do not] create a lightweight tag for each RCS revision
  25
  26 Config options:
  27         rcs.authorsFile         for --authors-file
  28         rcs.tagEachRev          for --tag-each-rev
  29
  30 EOM
  31 end
  32
  33 def not_found(arg)
  34         STDERR.puts "Could not find #{arg}"
  35 end
  36
  37 # returns a hash that maps usernames to author names & emails
  38 def load_authors_file(fn)
  39         hash = {}
  40         begin
  41                 File.open(File.expand_path fn) do |io|
  42                         io.each_line do |line|
  43                                 uname, author = line.split('=', 2)
  44                                 uname.strip!
  45                                 author.strip!
  46                                 STDERR.puts "Username #{uname} redefined to #{author}" if hash.has_key? uname
  47                                 hash[uname] = author
  48                         end
  49                 end
  50         rescue
  51                 not_found(fn)
  52         end
  53         return hash
  54 end
  55
  56 class Time
  57         def Time.rcs(string)
  58                 fields = string.split('.')
  59                 raise ArgumentError, "wrong number of fields for RCS date #{string}" unless fields.length == 6
  60                 Time.utc(*fields)
  61         end
  62 end
  63
  64 module RCS
  65         # strip an optional final ;
  66         def RCS.clean(arg)
  67                 arg.chomp(';')
  68         end
  69
  70         # strip the first and last @, and de-double @@s
  71         def RCS.sanitize(arg)
  72                 case arg
  73                 when Array
  74                         ret = arg.dup
  75                         raise 'malformed first line' unless ret.first[0,1] == '@'
  76                         raise 'malformed last line' unless ret.last[-1,1] == '@'
  77                         ret.first.sub!(/^@/,'')
  78                         ret.last.sub!(/@$/,'')
  79                         ret.map { |l| l.gsub('@@','@') }
  80                 when String
  81                         arg.chomp('@').sub(/^@/,'').gsub('@@','@')
  82                 else
  83                         raise
  84                 end
  85         end
  86
  87         # clean and sanitize
  88         def RCS.at_clean(arg)
  89                 RCS.sanitize RCS.clean(arg)
  90         end
  91
  92         def RCS.blob(arg)
  93                 arg.gsub('.', '0') + ('90'*5)
  94         end
  95
  96         def RCS.commit(arg)
  97                 arg.gsub('.', '0') + ('09'*5)
  98         end
  99
 100         class File
 101                 attr_accessor :head, :comment, :desc, :revision
 102                 def initialize(fname)
 103                         @fname = fname.dup
 104                         @head = nil
 105                         @comment = nil
 106                         @desc = []
 107                         @revision = Hash.new { |h, r| h[r] = Revision.new(r) }
 108                 end
 109
 110                 def has_revision?(rev)
 111                         @revision.has_key?(rev) and not @revision[rev].author.nil?
 112                 end
 113
 114                 def export_commits(opts={})
 115                         counter = 0
 116                         exported = []
 117                         until @revision.empty?
 118                                 counter += 1
 119
 120                                 # a string sort is a very good candidate for
 121                                 # export order, getting a miss only for
 122                                 # multi-digit revision components
 123                                 keys = @revision.keys.sort
 124
 125                                 STDERR.puts "commit export loop ##{counter}"
 126                                 STDERR.puts "\t#{exported.length} commits exported so far: #{exported.join(', ')}" unless exported.empty?
 127                                 STDERR.puts "\t#{keys.size} to export: #{keys.join(', ')}"
 128
 129                                 keys.each do |key|
 130                                         rev = @revision[key]
 131                                         # the parent commit is rev.next if we're on the
 132                                         # master branch (rev.branch is nil) or
 133                                         # rev.diff_base otherwise
 134                                         from = rev.branch.nil? ? rev.next : rev.diff_base
 135                                         # A commit can only be exported if it has no
 136                                         # parent, or if the parent has been exported
 137                                         # already. Skip this commit otherwise
 138                                         if from and not exported.include? from
 139                                                 next
 140                                         end
 141
 142                                         branch = rev.branch || 'master'
 143                                         author = opts[:authors][rev.author] || "#{rev.author} <empty>"
 144                                         date = "#{rev.date.tv_sec} +0000"
 145                                         log = rev.log.to_s
 146
 147                                         puts "commit refs/heads/#{branch}"
 148                                         puts "mark :#{RCS.commit key}"
 149                                         puts "committer #{author} #{date}"
 150                                         puts "data #{log.length}"
 151                                         puts log unless log.empty?
 152                                         puts "from :#{RCS.commit from}" if rev.branch_point
 153                                         puts "M 644 :#{RCS.blob key} #{@fname}"
 154
 155                                         # TODO FIXME this *should* be safe, in
 156                                         # that it should not unduly move
 157                                         # branches back in time, but I'm not
 158                                         # 100% sure ...
 159                                         rev.branches.each do |sym|
 160                                                 puts "reset refs/heads/#{sym}"
 161                                                 puts "from :#{RCS.commit key}"
 162                                         end
 163                                         rev.symbols.each do |sym|
 164                                                 puts "reset refs/tags/#{sym}"
 165                                                 puts "from :#{RCS.commit key}"
 166                                         end
 167                                         if opts[:tag_each_rev]
 168                                                 puts "reset refs/tags/#{key}"
 169                                                 puts "from :#{RCS.commit key}"
 170                                         end
 171
 172                                         exported.push key
 173                                 end
 174                                 exported.each { |k| @revision.delete(k) }
 175                         end
 176                 end
 177         end
 178
 179         class Revision
 180                 attr_accessor :rev, :author, :date, :state, :next
 181                 attr_accessor :branches, :log, :text, :symbols
 182                 attr_accessor :branch, :diff_base, :branch_point
 183                 def initialize(rev)
 184                         @rev = rev
 185                         @author = nil
 186                         @date = nil
 187                         @state = nil
 188                         @next = nil
 189                         @branches = []
 190                         @branch = nil
 191                         @branch_point = nil
 192                         @diff_base = nil
 193                         @log = []
 194                         @text = []
 195                         @symbols = []
 196                 end
 197
 198                 def date=(str)
 199                         @date = Time.rcs(str)
 200                 end
 201
 202                 def blob
 203                         str = @text.join('')
 204                         ret = "blob\nmark :#{RCS.blob @rev}\ndata #{str.length}\n#{str}\n"
 205                         ret
 206                 end
 207         end
 208
 209         def RCS.parse(fname, rcsfile, opts={})
 210                 rcs = RCS::File.new(fname)
 211
 212                 ::File.open(rcsfile, 'r') do |file|
 213                         status = [:basic]
 214                         rev = nil
 215                         lines = []
 216                         difflines = []
 217                         file.each_line do |line|
 218                                 case status.last
 219                                 when :basic
 220                                         command, args = line.split($;,2)
 221                                         next if command.empty?
 222
 223                                         case command
 224                                         when 'head'
 225                                                 rcs.head = RCS.clean(args.chomp)
 226                                         when 'symbols'
 227                                                 status.push :symbols
 228                                         when 'comment'
 229                                                 rcs.comment = RCS.at_clean(args.chomp)
 230                                         when /^[0-9.]+$/
 231                                                 rev = command.dup
 232                                                 if rcs.has_revision?(rev)
 233                                                         status.push :revision_data
 234                                                 else
 235                                                         status.push :new_revision
 236                                                 end
 237                                         when 'desc'
 238                                                 status.push :desc
 239                                                 lines.clear
 240                                                 status.push :read_lines
 241                                         else
 242                                                 STDERR.puts "Skipping unhandled command #{command.inspect}"
 243                                         end
 244                                 when :symbols
 245                                         sym, rev = line.strip.split(':',2);
 246                                         status.pop if rev.chomp!(';')
 247                                         rcs.revision[rev].symbols << sym
 248                                 when :desc
 249                                         rcs.desc.replace lines.dup
 250                                         status.pop
 251                                 when :read_lines
 252                                         # we sanitize lines as we read them
 253
 254                                         actual_line = line.dup
 255
 256                                         # the first line must begin with a @, which we strip
 257                                         if lines.empty?
 258                                                 ats = line.match(/^@+/)
 259                                                 raise 'malformed line' unless ats
 260                                                 actual_line.replace line.sub(/^@/,'')
 261                                         end
 262
 263                                         # if the line ends with an ODD number of @, it's the
 264                                         # last line -- we work on actual_line so that content
 265                                         # such as @\n or @ work correctly (they would be
 266                                         # encoded respectively as ['@@@\n','@\n'] and
 267                                         # ['@@@@\n']
 268                                         ats = actual_line.chomp.match(/@+$/)
 269                                         if nomore = (ats && Regexp.last_match(0).length.odd?)
 270                                                 actual_line.replace actual_line.chomp.sub(/@$/,'')
 271                                         end
 272                                         lines << actual_line.gsub('@@','@')
 273                                         if nomore
 274                                                 status.pop
 275                                                 redo
 276                                         end
 277                                 when :new_revision
 278                                         case line.chomp
 279                                         when /^date\s+(\S+);\s+author\s+(\S+);\sstate\s(\S+);$/
 280                                                 rcs.revision[rev].date = $1
 281                                                 rcs.revision[rev].author = $2
 282                                                 rcs.revision[rev].state = $3
 283                                         when 'branches'
 284                                                 status.push :branches
 285                                         when 'branches;'
 286                                                 next
 287                                         when /^next\s+(\S+)?;$/
 288                                                 nxt = rcs.revision[rev].next = $1
 289                                                 next unless nxt
 290                                                 raise "multiple diff_bases for #{nxt}" unless rcs.revision[nxt].diff_base.nil?
 291                                                 rcs.revision[nxt].diff_base = rev
 292                                                 rcs.revision[nxt].branch = rcs.revision[rev].branch
 293                                         else
 294                                                 status.pop
 295                                         end
 296                                 when :branches
 297                                         candidate = line.split(';',2)
 298                                         branch = candidate.first.strip
 299                                         rcs.revision[rev].branches.push branch
 300                                         raise "multiple diff_bases for #{branch}" unless rcs.revision[branch].diff_base.nil?
 301                                         rcs.revision[branch].diff_base = rev
 302                                         # we drop the last number from the branch name
 303                                         rcs.revision[branch].branch = branch.sub(/\.\d+$/,'.x')
 304                                         rcs.revision[branch].branch_point = rev
 305                                         status.pop if candidate.length > 1
 306                                 when :revision_data
 307                                         case line.chomp
 308                                         when 'log'
 309                                                 status.push :log
 310                                                 lines.clear
 311                                                 status.push :read_lines
 312                                         when 'text'
 313                                                 if rev == rcs.head
 314                                                         status.push :head
 315                                                 else
 316                                                         status.push :diff
 317                                                 end
 318                                                 lines.clear
 319                                                 status.push :read_lines
 320                                         else
 321                                                 status.pop
 322                                         end
 323                                 when :log
 324                                         rcs.revision[rev].log.replace lines.dup
 325                                         status.pop
 326                                 when :head
 327                                         rcs.revision[rev].text.replace lines.dup
 328                                         puts rcs.revision[rev].blob
 329                                         status.pop
 330                                 when :diff
 331                                         difflines.replace lines.dup
 332                                         difflines.pop if difflines.last.empty?
 333                                         base = rcs.revision[rev].diff_base
 334                                         unless rcs.revision[base].text
 335                                                 pp rcs
 336                                                 puts rev, base
 337                                                 raise 'no diff base!'
 338                                         end
 339                                         # deep copy
 340                                         buffer = []
 341                                         rcs.revision[base].text.each { |l| buffer << [l.dup] }
 342
 343                                         adding = false
 344                                         index = nil
 345                                         count = nil
 346
 347                                         while l = difflines.shift
 348                                                 if adding
 349                                                         raise 'negative index during insertion' if index < 0
 350                                                         raise 'negative count during insertion' if count < 0
 351                                                         adding << l
 352                                                         count -= 1
 353                                                         # collected all the lines, put the before
 354                                                         unless count > 0
 355                                                                 buffer[index].unshift *adding
 356                                                                 adding = false
 357                                                         end
 358                                                         next
 359                                                 end
 360
 361                                                 l.chomp!
 362                                                 raise 'malformed diff' unless l =~ /^([ad])(\d+) (\d+)$/
 363                                                 diff_cmd = $1.intern
 364                                                 index = $2.to_i
 365                                                 count = $3.to_i
 366                                                 case diff_cmd
 367                                                 when :d
 368                                                         # for deletion, index 1 is the first index, so the Ruby
 369                                                         # index is one less than the diff one
 370                                                         index -= 1
 371                                                         # we replace them with empty string so that 'a' commands
 372                                                         # referring to the same line work properly
 373                                                         while count > 0
 374                                                                 buffer[index].clear
 375                                                                 index += 1
 376                                                                 count -= 1
 377                                                         end
 378                                                 when :a
 379                                                         # addition will prepend the appropriate lines
 380                                                         # to the given index, and in this case Ruby
 381                                                         # and diff indices are the same
 382                                                         adding = []
 383                                                 end
 384                                         end
 385
 386                                         # turn the buffer into an array of lines, deleting the empty ones
 387                                         buffer.delete_if { |l| l.empty? }
 388                                         buffer.flatten!
 389
 390                                         rcs.revision[rev].text = buffer
 391                                         puts rcs.revision[rev].blob
 392                                         status.pop
 393                                 else
 394                                         STDERR.puts "Unknown status #{status.last}"
 395                                         exit 1
 396                                 end
 397                         end
 398                 end
 399
 400                 # clean up the symbols/branches: look for revisions that have
 401                 # one or more symbols but no dates, and make them into
 402                 # branches, pointing to the highest commit with that key
 403                 branches = []
 404                 keys = rcs.revision.keys
 405                 rcs.revision.each do |key, rev|
 406                         if rev.date.nil? and not rev.symbols.empty?
 407                                 top = keys.select { |k| k.match(/^#{key}\./) }.sort.last
 408                                 tr = rcs.revision[top]
 409                                 raise "unhandled complex branch structure met: #{rev.inspect} refers #{tr.inspect}" if tr.date.nil?
 410                                 tr.branches |= rev.symbols
 411                                 branches << key
 412                         end
 413                 end
 414                 branches.each { |k| rcs.revision.delete k }
 415
 416                 # export the commits
 417                 rcs.export_commits(opts)
 418         end
 419 end
 420
 421 require 'getoptlong'
 422
 423 opts = GetoptLong.new(
 424         # Authors file, like git-svn and git-cvsimport, more than one can be
 425         # specified
 426         ['--authors-file', '-A', GetoptLong::REQUIRED_ARGUMENT],
 427         # RCS file suffix, like RCS
 428         ['--rcs-suffixes', '-x', GetoptLong::REQUIRED_ARGUMENT],
 429         # tag each revision?
 430         ['--tag-each-rev', GetoptLong::NO_ARGUMENT],
 431         ['--no-tag-each-rev', GetoptLong::NO_ARGUMENT],
 432         ['--help', '-h', '-?', GetoptLong::NO_ARGUMENT]
 433 )
 434
 435 # We read options in order, but they apply to all passed parameters.
 436 # TODO maybe they should only apply to the following, unless there's only one
 437 # file?
 438 opts.ordering = GetoptLong::RETURN_IN_ORDER
 439
 440 file_list = []
 441 parse_options = {
 442         :authors => Hash.new,
 443 }
 444
 445 # Read config options
 446 `git config --get-all rcs.authorsfile`.each_line do |fn|
 447         parse_options[:authors].merge! load_authors_file(fn.chomp)
 448 end
 449
 450 parse_options[:tag_each_rev] = (
 451         `git config --bool rcs.tageachrev`.chomp == 'true'
 452 ) ? true : false
 453
 454 opts.each do |opt, arg|
 455         case opt
 456         when '--authors-file'
 457                 authors = load_authors_file(arg)
 458                 redef = parse_options[:authors].keys & authors.keys
 459                 STDERR.puts "Authors file #{arg} redefines #{redef.join(', ')}" unless redef.empty?
 460                 parse_options[:authors].merge!(authors)
 461         when '--rcs-suffixes'
 462                 # TODO
 463         when '--tag-each-rev'
 464                 parse_options[:tag_each_rev] = true
 465         when '--no-tag-each-rev'
 466                 # this is the default, which is fine since the missing key
 467                 # (default) returns nil which is false in Ruby
 468                 parse_options[:tag_each_rev] = false
 469         when ''
 470                 file_list << arg
 471         when '--help'
 472                 usage
 473                 exit
 474         end
 475 end
 476
 477 require 'etc'
 478
 479 user = Etc.getlogin || ENV['USER']
 480
 481 # steal username/email data from other init files that may contain the
 482 # information
 483 def steal_username
 484         [
 485                 # the user's .hgrc file for a username field
 486                 ['~/.hgrc',   /^\s*username\s*=\s*(["'])?(.*)\1$/,       2],
 487                 # the user's .(g)vimrc for a changelog_username setting
 488                 ['~/.vimrc',  /changelog_username\s*=\s*(["'])?(.*)\1$/, 2],
 489                 ['~/.gvimrc', /changelog_username\s*=\s*(["'])?(.*)\1$/, 2],
 490                 []
 491         ].each do |fn, rx, idx|
 492                 file = File.expand_path fn
 493                 if File.readable?(file) and File.read(file) =~ rx
 494                         parse_options[:authors][user] = Regexp.last_match(idx).strip
 495                         break
 496                 end
 497         end
 498 end
 499
 500 if user and not user.empty? and not parse_options[:authors].has_key?(user)
 501         name = ENV['GIT_AUTHOR_NAME'] || ''
 502         name.replace(`git config user.name`.chomp) if name.empty?
 503         name.replace(Etc.getpwnam(user).gecos) if name.empty?
 504
 505         if name.empty?
 506                 # couldn't find a name, try to steal data from other sources
 507                 steal_username
 508         else
 509                 # if we found a name, try to find an email too
 510                 email = ENV['GIT_AUTHOR_EMAIL'] || ''
 511                 email.replace(`git config user.email`.chomp) if email.empty?
 512
 513                 if email.empty?
 514                         # couldn't find an email, try to steal data too
 515                         steal_username
 516                 else
 517                         # we got both a name and email, fill the info
 518                         parse_options[:authors][user] = "#{name} <#{email}>"
 519                 end
 520         end
 521 end
 522
 523 if file_list.empty?
 524         usage
 525         exit 1
 526 end
 527
 528 SFX = ',v'
 529
 530 status = 0
 531
 532 file_list.each do |arg|
 533         if arg[-2,2] == SFX
 534                 if File.exists? arg
 535                         rcsfile = arg.dup
 536                 else
 537                         not_found "RCS file #{arg}"
 538                         status |= 1
 539                 end
 540                 filename = File.basename(arg, SFX)
 541         else
 542                 filename = File.basename(arg)
 543                 path = File.dirname(arg)
 544                 rcsfile = File.join(path, 'RCS', filename) + SFX
 545                 unless File.exists? rcsfile
 546                         rcsfile.replace File.join(path, filename) + SFX
 547                         unless File.exists? rcsfile
 548                                 not_found "RCS file for #{filename} in #{path}"
 549                         end
 550                 end
 551         end
 552
 553         RCS.parse(filename, rcsfile, parse_options)
 554 end
 555
 556 exit status