From a7cdc79ed3f3acf8680bead604aba9eb3a6984fb Mon Sep 17 00:00:00 2001
From: VVSiz <i.4m.l33t@yandex.ru>
Date: Mon, 7 Mar 2011 13:59:57 +0100
Subject: [PATCH] dsl2txt script added

---
 dsl2txt.rb           | 81 ++++++++++++++++++++++++++++++++++++++++++++
 lib/Card.rb          |  2 +-
 lib/Dictionary.rb    | 21 ++++++++++++
 lib/TxtDictionary.rb | 32 +++++++++++++++++
 4 files changed, 135 insertions(+), 1 deletion(-)
 create mode 100644 dsl2txt.rb
 create mode 100644 lib/TxtDictionary.rb

diff --git a/dsl2txt.rb b/dsl2txt.rb
new file mode 100644
index 0000000..9ee347a
--- /dev/null
+++ b/dsl2txt.rb
@@ -0,0 +1,81 @@
+# TODO:
+# 1. check for key, value, desc validity
+# 2. check for empty cards
+
+require 'optparse'
+require 'fileutils'
+
+require File.expand_path('../lib/Dictionary', __FILE__)
+require File.expand_path('../lib/Card', __FILE__)
+require File.expand_path('../lib/TxtDictionary', __FILE__)
+
+$IN = ""
+$OUT = ""
+$DOT = "*"
+$DESC_DOT = "-"
+
+$separotor = nil
+opts = OptionParser.new
+opts.on("-i FILE", "--in FILE", "input file to convert", String) {|val| $IN = val }
+opts.on("-o DIR", "--out DIR", "directory to extract txt files", String) { |val|
+  if ($OUT.empty?)
+    $OUT = val
+  else
+    $stderr.puts "Error: output file can be defined only once!\n\n"
+    $stderr.puts opts
+    exit
+  end
+}
+opts.on("-s", "start each line with a separator") { $separator = $DOT }
+opts.on_tail("-h", "--help", "Show this message") do
+  $stderr.puts opts
+  exit
+end
+rest = opts.parse(ARGV)
+
+if ($IN.size == 0 || $OUT.size == 0)
+  $stderr.puts opts
+  exit
+end
+
+# load the DSL dictionary
+d = nil
+File.open($IN, 'rb') { |f|
+  d = Dictionary.load_from_dsl(f)
+}
+
+if File.exist?($OUT)
+  $stderr.puts "ERROR: Output directory already exist: '#{$OUT}'"
+  # TODO: must be enabled
+  # exit
+  
+  # back = $OUT + ".bak"
+  # FileUtils.copy($OUT, back)
+  # $stderr.puts "Made backup copy: #{back}"
+  # File.open($OUT, 'rb') { |f|
+  #  d = Dictionary.load_from_dsl(f)
+  #}
+else
+  $stderr.puts "Creating output directory: '#{$OUT}'"
+  Dir.mkdir($OUT)
+end
+
+[].each { |file_name|
+  desc = ""
+  File.foreach(file_name) { |line|
+    if line =~ /^(?:\xEF\xBB\xBF)?#\s*[Dd]escription\s*(.*)$/
+      desc = $1.strip
+    elsif line =~ /^\s*$/
+      # ignore
+    elsif line =~ /^(.*?):(.*)$/
+      key, value = $1, $2
+      d.add(key, value, desc)
+    end
+  }
+}
+
+d.get_txt_dicts.each { |dict|
+  dict.print_out("CON")
+  dict.extract_to_dir($OUT)
+}
+
diff --git a/lib/Card.rb b/lib/Card.rb
index a07e84e..44f462b 100644
--- a/lib/Card.rb
+++ b/lib/Card.rb
@@ -1,5 +1,5 @@
 class Card
-  attr_accessor :headword
+  attr_accessor :headword, :entries
   def initialize(headword)
     @headword = headword
     @entries = {}
diff --git a/lib/Dictionary.rb b/lib/Dictionary.rb
index e924ff1..47a6fd5 100644
--- a/lib/Dictionary.rb
+++ b/lib/Dictionary.rb
@@ -90,4 +90,25 @@ def print_out(out_name)
     $stderr.puts "File #{out_name} written...."
     $stderr.puts "Total number of headwords: #{@cards.size}"
   end
+  
+  def each_card
+    @cards.values.sort.each { |card|
+      yield card
+    }
+  end
+  
+  def get_txt_dicts
+    txt_dicts = {}
+
+    each_card { |card|
+      hwd = card.headword
+      card.entries.each { |trn, descs|
+        descs.each { |desc|
+          (txt_dicts[desc] ||= TxtDictionary.new(desc)).add(hwd, trn)
+        }
+      }
+    }
+
+    txt_dicts.values
+  end
 end
diff --git a/lib/TxtDictionary.rb b/lib/TxtDictionary.rb
new file mode 100644
index 0000000..924b0ab
--- /dev/null
+++ b/lib/TxtDictionary.rb
@@ -0,0 +1,32 @@
+class TxtDictionary
+  def initialize(desc)
+    @desc = desc
+    @data = []
+  end
+
+  def add(key, trn)
+    @data << [key, trn]
+  end
+
+  def extract_to_dir(dir_name)
+    file = dir_name + "/" + desc_to_filename + ".txt"
+    if (File.exist?(file))
+      $stderr.puts "ERROR: file already exist: '#{file}'"
+      exit
+    end
+    print_out(file)
+  end
+
+  def print_out(out_name)
+    out = File.open(out_name, 'w')
+    out.puts "#description #{@desc}"
+    out.puts ""
+    @data.each { |line|
+      out.puts "#{line[0]}:#{line[1]}"
+    }
+  end
+
+  def desc_to_filename
+    @desc.gsub('"', '').gsub(/\s\\#/, '_')
+  end
+end