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