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

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