fixup-migrate-gettext.rb

   1 #!/usr/bin/ruby
   2 # coding: utf-8
   3 class PoStr
   4   Sections = %w(Headers Source Translated)
   5   attr_accessor :status, :src, :srcclean, :dest, :prev, :prevclean, :type, :location
   6   def initialize(text)
   7     section = 0
   8     @status = ''
   9     @src = []
  10     @dest = []
  11     @prev = []
  12
  13     text.split(/\n/).each do |lin|
  14       # Header
  15       if section == 0
  16         if lin =~ /^#(.) ([^:]*)(?::(.*))?/
  17           type = $1
  18           field = $2
  19           value = $3
  20
  21           @status = 'fuzzy' if type == ',' and field == 'fuzzy'
  22           @type = value if type == '.' and field == 'type'
  23           @location = {:file => field, :line => value} if type == ':'
  24           if type == '|'
  25             #print lin, "\n"
  26             if lin =~ /^\#\| msgid "(.*)"$/
  27               #print "first '%s'\n" % $1
  28               @prev << $1
  29             elsif lin =~ /^\#\| "(.*)"$/
  30               @prev << $1
  31             end
  32           end
  33         elsif lin =~ /^msgid "(.*)"/
  34           section += 1
  35           @src << $1
  36         else
  37           boom(section, lin, text)
  38         end
  39
  40       # Source text
  41       elsif section == 1
  42         if lin =~ /^"(.*)"$/
  43           @src << $1
  44         elsif lin =~ /^msgstr "(.*)"/
  45           section += 1
  46           @dest << $1
  47         else
  48           boom(section, lin, text)
  49         end
  50
  51       # Translated text
  52       else
  53         if lin =~ /^"(.*)"$/
  54           @dest << $1
  55         else
  56           boom(section, lin, text)
  57         end
  58       end
  59     end
  60   end
  61
  62   def boom(section, line, text)
  63     raise RuntimeError, "Unexpected string in %s:\n %s\n\nFull text: %s" %
  64                         [Sections[section], line, text]
  65   end
  66
  67   def fuzzy?
  68     @status == 'fuzzy'
  69   end
  70
  71   def srcstr()
  72     return @src.join("")
  73   end
  74
  75   def deststr()
  76     return @dest.join("")
  77   end
  78
  79   def prevstr()
  80     return @prev.join("")
  81   end
  82
  83   def output()
  84     # No use outputting empty translations
  85     if '' == srcstr() && '' == deststr()
  86       return
  87     end
  88     if '' != @type
  89       print "#. type:%s\n" % [@type]
  90     end
  91     if fuzzy?
  92       print "#, fuzzy\n"
  93       if '' != prevstr()
  94         @prev.each_with_index do |lin, idx|
  95           if idx == 0
  96             print "#| msgid \"%s\"\n" % lin
  97           else
  98             print "#| \"%s\"\n" % lin
  99           end
 100         end
 101       end
 102     end
 103     @src.each_with_index do |lin, idx|
 104       if idx == 0
 105         print "msgid \"%s\"\n" % lin
 106       else
 107         print "\"%s\"\n" % lin
 108       end
 109     end
 110     @dest.each_with_index do |lin, idx|
 111       if idx == 0
 112         print "msgstr \"%s\"\n" % lin
 113       else
 114         print "\"%s\"\n" % lin
 115       end
 116     end
 117     print "\n"
 118   end
 119 end
 120
 121 raise ArgumentError, 'Source file not specified' if ARGV.size != 1
 122 file = ARGV[0]
 123 strings = File.open(file,'r').read.split(/\n\n/)[1..-1].map {|str| PoStr.new(str)}
 124
 125 #print strings
 126
 127 c = 0
 128 strings.each do |entry|
 129   if entry.fuzzy?
 130     # Ignore whitespace changes between prev and src
 131     entry.prevclean = entry.prevstr().gsub(/ +/, ' ')
 132     entry.srcclean = entry.srcstr().gsub(/ +/, ' ')
 133
 134     if entry.prevclean == entry.srcclean
 135       entry.status = ''
 136       #print "clear fuzzy (space)\n"
 137       c = c + 1
 138       next
 139     end
 140
 141     # Rewrite title strings, which lost '#' at the front
 142     if entry.prevclean.gsub(/^#+ +/, '') == entry.srcclean
 143       entry.status = ''
 144       # FIXME
 145       if entry.dest[0].gsub!(/^#+ +/, '') ||
 146          ('' == entry.dest[0] && entry.dest[1].gsub!(/^#+ +/, ''))
 147         #print "cleared fuzzy (title)\n"
 148         c = c + 1
 149       end
 150     end
 151
 152     # Rewrite footnotes to use "<placeholder type=\"footnote\" id=\"0\"/>"
 153     if entry.srcclean =~ /<placeholder type=\\"footnote\\" id=\\"0\\"\/>/
 154       #print "found footnote\n"
 155       p = entry.prevclean
 156       p.sub!(/([a-z]\.["”]?)(\d+)(\s)/, "\\1<placeholder type=\\\"footnote\\\" id=\\\"0\\\"\/>\\3")
 157       p.sub!(/([a-z]\.["”]?)(\d+)$/, "\\1<placeholder type=\\\"footnote\\\" id=\\\"0\\\"\/>")
 158       #print p, "\n"
 159       #print entry.src, "\n"
 160       if p == entry.srcclean
 161         replaced = false
 162         entry.dest.each do |part|
 163           if part.sub!(/([a-z]\.["”»]?)(\d+)(\s)/, "\\1<placeholder type=\\\"footnote\\\" id=\\\"0\\\"\/>\\3")
 164             replaced = true
 165           end
 166         end
 167         entry.dest.each do |part|
 168           if part.sub!(/([a-z]\.["”»]?)(\d+)$/, "\\1<placeholder type=\\\"footnote\\\" id=\\\"0\\\"\/>")
 169             replaced = true
 170           end
 171         end
 172         if replaced
 173           #print "cleared fuzzy (footnote)\n"
 174           entry.status = ''
 175           c = c + 1
 176         end
 177       end
 178     end
 179   end
 180   entry.output
 181 end
 182 print "# Would clear %d fuzzy\n" % [c]