Start on code to clear bogus fuzzies.

[text-madewithcc.git] / fixup-migrate-gettext.rb

diff --git a/fixup-migrate-gettext.rb b/fixup-migrate-gettext.rb
old mode 100644 (file)
new mode 100755 (executable)
index 9bb7ec8..66785d3
--- a/fixup-migrate-gettext.rb
+++ b/fixup-migrate-gettext.rb
@@ -1,12 +1,14 @@
 #!/usr/bin/ruby
+# coding: utf-8
 class PoStr
   Sections = %w(Headers Source Translated)
-  attr_accessor :status, :src, :dest, :type, :location
+  attr_accessor :status, :src, :dest, :prev, :type, :location
   def initialize(text)
     section = 0
     @status = ''
-    @src = []
-    @dest = []
+    @src = ''
+    @dest = ''
+    @prev = ''
 
     text.split(/\n/).each do |lin|
       # Header
@@ -19,7 +21,14 @@ class PoStr
           @status = 'fuzzy' if type == ',' and field == 'fuzzy'
           @type = value if type == '.' and field == 'type'
           @location = {:file => field, :line => value} if type == ':'
-        elsif lin =~ /^msgid (".*")/
+          if type == '|'
+            if field =~ /^msgid "(.*)"/
+              @prev << $1
+            elsif field =~ /^"(.*)"$/
+              @prev << $1
+            end
+          end
+        elsif lin =~ /^msgid "(.*)"/
           section += 1
           @src << $1
         else
@@ -28,9 +37,9 @@ class PoStr
 
       # Source text
       elsif section == 1
-        if lin =~ /^".*"$/
-          @src << lin
-        elsif lin =~ /^msgstr (".*")/
+        if lin =~ /^"(.*)"$/
+          @src << $1
+        elsif lin =~ /^msgstr "(.*)"/
           section += 1
           @dest << $1
         else
@@ -39,8 +48,8 @@ class PoStr
 
       # Translated text
       else
-        if lin =~ /^".*"$/
-          @dest << lin
+        if lin =~ /^"(.*)"$/
+          @dest << $1
         else
           boom(section, lin, text)
         end
@@ -56,7 +65,65 @@ class PoStr
   def fuzzy?
     @status == 'fuzzy'
   end
+
+  def output()
+    # FIXME rewrite to output proper entries
+    if '' != @type
+      print "#. type: %s\n" % [@type]
+    end
+    if fuzzy?
+      print "#, fuzzy\n"
+      if '' != @prev
+        print "#| msgid \"%s\"\n" % [@prev]
+      end
+    end
+    print "msgid \"%s\"\n" % [@src]
+    print "msgstr \"%s\"\n" % [@dest]
+    print "\n\n"
+  end
 end
 
-file = 'po/es/mwcc.po'
+raise ArgumentError, 'Source file not specified' if ARGV.size != 1
+file = ARGV[0]
 strings = File.open(file,'r').read.split(/\n\n/)[1..-1].map {|str| PoStr.new(str)}
+
+print strings
+
+c = 0
+strings.each do |entry|
+  if entry.fuzzy?
+    # Ignore whitespace changes between prev and src
+    if entry.prev.gsub(/ +/, ' ') == entry.src.gsub(/ +/, ' ')
+      entry.status = ''
+      print "clear fuzzy (space)\n"
+      c = c + 1
+    end
+
+    # Rewrite title strings, which lost '#' at the front
+    if entry.prev.gsub(/^#+ +/, '') == entry.src
+      entry.status = ''
+      entry.dest.gsub!(/^#+ +/, '')
+      print "cleared fuzzy (title)\n"
+      c = c + 1
+    end
+
+    # Rewrite footnotes to use "<placeholder type=\"footnote\" id=\"0\"/>"
+    if entry.src =~ /<placeholder type=\\"footnote\\" id=\\"0\\"\/>/
+      print "found footnote\n"
+      p = entry.prev
+      p.sub!(/([a-z]\.["”]?)(\d+)(\s)/, "\\1<placeholder type=\\\"footnote\\\" id=\\\"0\\\"\/>\\3")
+      p.sub!(/([a-z]\.["”]?)(\d+)$/, "\\1<placeholder type=\\\"footnote\\\" id=\\\"0\\\"\/>")
+      print p, "\n"
+      print entry.src, "\n"
+      if p == entry.src
+        entry.status = ''
+        entry.dest.sub!(/([a-z]\.["”»]?)(\d+)(\s)/, "\\1<placeholder type=\\\"footnote\\\" id=\\\"0\\\"\/>\\3")
+        entry.dest.sub!(/([a-z]\.["”»]?)(\d+)$/, "\\1<placeholder type=\\\"footnote\\\" id=\\\"0\\\"\/>")
+        print "cleared fuzzy (footnote)\n"
+        c = c + 1
+      end
+    end
+  end
+  entry.output
+end
+print "Would clear %d fuzzy\n" % [c]