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 @location
  92       print "#: %s:%s\n" % [@location[:file], @location[:line]]
  93     end
  94     if fuzzy?
  95       print "#, fuzzy\n"
  96       if '' != prevstr()
  97         @prev.each_with_index do |lin, idx|
  98           if idx == 0
  99             print "#| msgid \"%s\"\n" % lin
 100           else
 101             print "#| \"%s\"\n" % lin
 102           end
 103         end
 104       end
 105     end
 106     @src.each_with_index do |lin, idx|
 107       if idx == 0
 108         print "msgid \"%s\"\n" % lin
 109       else
 110         print "\"%s\"\n" % lin
 111       end
 112     end
 113     @dest.each_with_index do |lin, idx|
 114       if idx == 0
 115         print "msgstr \"%s\"\n" % lin
 116       else
 117         print "\"%s\"\n" % lin
 118       end
 119     end
 120     print "\n"
 121   end
 122 end
 123
 124 raise ArgumentError, 'Source file not specified' if ARGV.size != 1
 125 file = ARGV[0]
 126 strings = File.open(file,'r').read.split(/\n\n/)[1..-1].map {|str| PoStr.new(str)}
 127
 128 #print strings
 129
 130 c = 0
 131 strings.each do |entry|
 132   # Ignore whitespace changes between prev and src
 133   entry.prevclean = entry.prevstr().gsub(/ +/, ' ')
 134   entry.srcclean = entry.srcstr().gsub(/ +/, ' ')
 135
 136   if entry.fuzzy?
 137     if entry.prevclean == entry.srcclean
 138       entry.status = ''
 139       #print "# clear fuzzy (space)\n"
 140       c = c + 1
 141       entry.output
 142       next
 143     end
 144
 145     # Rewrite title strings, which lost '#' at the front
 146     if entry.prevclean.gsub(/^#+ +/, '') == entry.srcclean
 147       # FIXME Not safe to clear fuzzy flag, might have bogus translation
 148       entry.status = ''
 149       if entry.dest[0].gsub!(/^#+ +/, '') ||
 150          ('' == entry.dest[0] && entry.dest[1].gsub!(/^#+ +/, ''))
 151         #print "cleared fuzzy (title)\n"
 152         c = c + 1
 153       end
 154     end
 155
 156     # Rewrite footnotes to use "<placeholder type=\"footnote\" id=\"0\"/>"
 157     if entry.srcclean =~ /<placeholder type=\\"footnote\\" id=\\"0\\"\/>/
 158       #print "found footnote\n"
 159       p = entry.prevclean
 160       p.sub!(/([a-z]\.["”]?)(\d+)(\s)/, "\\1<placeholder type=\\\"footnote\\\" id=\\\"0\\\"\/>\\3")
 161       p.sub!(/([a-z]\.["”]?)(\d+)$/, "\\1<placeholder type=\\\"footnote\\\" id=\\\"0\\\"\/>")
 162       #print p, "\n"
 163       #print entry.src, "\n"
 164       if p == entry.srcclean
 165         replaced = false
 166         entry.dest.each do |part|
 167           if part.sub!(/([a-z]\.["”»]?)(\d+)(\s)/, "\\1<placeholder type=\\\"footnote\\\" id=\\\"0\\\"\/>\\3")
 168             replaced = true
 169           end
 170         end
 171         entry.dest.each do |part|
 172           if part.sub!(/([a-z]\.["”»]?)(\d+)$/, "\\1<placeholder type=\\\"footnote\\\" id=\\\"0\\\"\/>")
 173             replaced = true
 174           end
 175         end
 176         if replaced
 177           #print "cleared fuzzy (footnote)\n"
 178           entry.status = ''
 179           c = c + 1
 180         end
 181       end
 182     end
 183   end
 184
 185   # Copy image references and single URLs unchanged, even for non-fuzzies
 186   if entry.prevclean =~ /!\[\]\(Pictures\// ||
 187      entry.srcclean =~ /^<ulink url=\\\".+\\\"\/>$/
 188     entry.dest = entry.src.clone()
 189     entry.status = ''
 190     c = c + 1
 191   end
 192
 193   entry.output
 194 end
 195 print "# Cleared %d fuzzy strings\n" % [c]