Ensure that transfer.dat lines have always the correct line length of 97

History.txt

@@ -1,6 +1,7 @@
 === 2.4.9 / 15.01.2018
 
 * Adapt to new link of epha atc.csv
+* Ensure that transfer.dat lines have always the correct line length of 97
 
 === 2.4.8 / 20.11.2017
 

lib/oddb2xml/builder.rb

@@ -1182,20 +1182,10 @@ def format_date(date_str, len=7)
       end
       date[0,len]
     end
-    def format_name(name)
-      if RUBY_VERSION.to_f < 1.9 # for multibyte chars length support
-        chars = name.scan(/./mu).to_a
-        diff  = DAT_LEN[:ABEZ] - chars.length
-        if diff > 0
-          chars += Array.new(diff, ' ')
-        elsif diff < 0
-          chars = chars[0,DAT_LEN[:ABEZ]]
-        end
-        chars.to_s
-      else
-        name = name[0,DAT_LEN[:ABEZ]] if name.length > DAT_LEN[:ABEZ]
-        "%-#{DAT_LEN[:ABEZ]}s" % name
-      end
+
+    # The migel name must be always 50 chars wide and in ISO 8859-1 format
+    def format_name(name, length)
+      ("%-#{length}s" % name)[0..length-1]
     end
     def build_dat
       prepare_articles
@@ -1248,7 +1238,12 @@ def build_dat
             price_exf     = sprintf('%06i', (pac[:prices][:exf_price][:price].to_f*100).to_i) if pac[:prices][:exf_price] and pac[:prices][:exf_price][:price]
             price_public  = sprintf('%06i', (pac[:prices][:pub_price][:price].to_f*100).to_i) if pac[:prices][:pub_price] and pac[:prices][:pub_price][:price]
           end
-          row << format_name(abez)
+          row << format_name(Oddb2xml.patch_some_utf8(abez), DAT_LEN[:ABEZ])
+          if price_exf.to_s.length > DAT_LEN[:PRMO] ||
+             price_public.to_s.length > DAT_LEN[:PRPU]
+            puts "Price exfactory #{price_exf} or public #{price_public} is too high to be added into transfer.dat"
+            break
+          end
           row << "%#{DAT_LEN[:PRMO]}s"  % (price_exf ? price_exf.to_s : ('0' * DAT_LEN[:PRMO]))
           row << "%#{DAT_LEN[:PRPU]}s"  % (price_public ? price_public.to_s : ('0' * DAT_LEN[:PRPU]))
           row << "%#{DAT_LEN[:CKZL]}s"  % if (@lppvs[ean])
@@ -1303,9 +1298,9 @@ def build_with_migel_dat
             obj[:desc_de].to_s + " " +
           (obj[:quantity] ? obj[:quantity] : '')
           ).gsub(/"/, '')
-          row << format_name(abez)
-          row << "%#{DAT_LEN[:PRMO]}s"  % ('0' * DAT_LEN[:PRMO])
-          row << "%#{DAT_LEN[:PRPU]}s"  % ('0' * DAT_LEN[:PRPU])
+          row << format_name(Oddb2xml.patch_some_utf8(abez), DAT_LEN[:ABEZ])
+          row << '0' * DAT_LEN[:PRMO]
+          row << '0' * DAT_LEN[:PRPU]
           row << "%#{DAT_LEN[:CKZL]}s"  % '3' # sl_entry and lppv
           row << "%#{DAT_LEN[:CLAG]}s"  % '0'
           row << "%#{DAT_LEN[:CBGG]}s"  % '0'

lib/oddb2xml/extractor.rb

@@ -470,7 +470,7 @@ def initialize(dat, extended = false)
     def to_hash
       data = {}
       while line = @io.gets
-        line = line.encode('utf-8').gsub("\u0089", "‰").gsub("\u0092", '’').gsub("\u0096", '-').chomp
+        line = Oddb2xml.patch_some_utf8(line)
         next if line =~ /(ad us\.* vet)|(\(vet\))/i
         if @@extended
           next unless line =~ /(\d{13})(\d{1})$/

lib/oddb2xml/util.rb

@@ -20,25 +20,24 @@ def Oddb2xml.calc_checksum(str)
   @atc_csv_origin = 'http://download.epha.ch/data/atc/atc.csv'
   @atc_csv_content = {}
 
+  def Oddb2xml.patch_some_utf8(line)
+    begin
+      line.gsub("\u0089", "‰").gsub("\u0092", '’').gsub("\u0096", '-').gsub("\u2013",'-').gsub("\u201D", '"').chomp
+    rescue => error
+      puts "#{error}: in #{line}"
+      line
+    end
+  end
+
   def Oddb2xml.convert_to_8859_1(line)
     begin
       # We want to ignore lines which are not really UTF-8 encoded
-      return line.encode('ISO-8859-1')
+        ausgabe = Oddb2xml.patch_some_utf8(line).encode('ISO-8859-1')
+        ausgabe.encode('ISO-8859-1')
     rescue => error
-      ausgabe = ''
-      0.upto(line.size-1).each do |idx|
-        begin
-          if line[idx].ord == 8211
-            ausgabe += '-'
-          else
-            ausgabe += line[idx].encode('ISO-8859-1')
-          end
-        rescue => error
-          puts "#{error}: in #{line} at #{idx}"
-        end
-      end
+      puts "#{error}: in #{line}"
+      require 'pry'; binding.pry
     end
-    ausgabe.encode('ISO-8859-1')
   end
 
   def Oddb2xml.add_epha_changes_for_ATC(iksnr, atc_code)