swissmedic-diff 0.1.5 → 0.1.6

data/Gemfile.lock

@@ -1,7 +1,10 @@
 PATH
   remote: .
   specs:
-    swissmedic-diff (0.1.4)
+    swissmedic-diff (0.1.5)
+      nokogiri
+      rubyXL
+      rubyzip
       spreadsheet
 
 GEM
@@ -10,19 +13,24 @@ GEM
     ansi (1.4.3)
     builder (3.2.2)
     hashie (2.0.5)
-    hoe (3.7.2)
+    hoe (3.7.4)
       rake (>= 0.8, < 11.0)
-    minitest (5.1.0)
+    mini_portile (0.5.2)
+    minitest (5.2.1)
     minitest-reporters (1.0.0)
       ansi
       builder
       minitest (>= 5.0)
       powerbar
+    nokogiri (1.6.0)
+      mini_portile (~> 0.5.0)
     powerbar (1.0.11)
       ansi (~> 1.4.0)
       hashie (>= 1.1.0)
-    rake (10.1.0)
+    rake (10.1.1)
     ruby-ole (1.2.11.7)
+    rubyXL (1.2.10)
+    rubyzip (0.9.9)
     spreadsheet (0.9.6)
       ruby-ole (>= 1.0)
 

data/History.txt

@@ -1,4 +1,11 @@
-=== 0.1.5 / 11.12.2010
+=== 0.1.6 / 14.01.2014
+
+* Added test/data/Packungen-2014.xlsx
+* Added missing dependencies for rubyXL
+* Moved VERSION to separate file
+* Use rubyXL for xslx files
+
+=== 0.1.5 / 11.12.2013
 
 * Updated tests as we no longer raise an error
 * Update to spreadsheet 0.9.6

data/Manifest.txt

@@ -9,9 +9,11 @@ README.txt
 Rakefile
 bin/swissmedic-diff
 lib/swissmedic-diff.rb
+lib/version.rb
 setup.rb
 swissmedic-diff.gemspec
 test/data/Packungen-2013.10.14.xls
+test/data/Packungen-2014.xlsx
 test/data/Packungen.older.xls
 test/data/Packungen.xls
 test/data/Packungen_error_column.xls

data/lib/swissmedic-diff.rb

@@ -4,6 +4,7 @@
 
 require 'ostruct'
 require 'spreadsheet'
+require 'rubyXL'
 
 #= diff command (compare two xls fles) for swissmedic xls file.
 #
@@ -17,8 +18,6 @@ require 'spreadsheet'
 #License::   GPLv2.0 Compliance 
 #Source::    http://scm.ywesee.com/?p=swissmedic-diff/.git;a=summary
 class SwissmedicDiff
-  VERSION = '0.1.5'
-
   module Diff
     COLUMNS = [ :iksnr, :seqnr, :name_base, :company, 
                 :index_therapeuticus, :atc_class, :production_science,
@@ -97,7 +96,11 @@ class SwissmedicDiff
       @diff.changes = changes = {}
       @diff.newest_rows = newest_rows
       Spreadsheet.client_encoding = 'UTF-8'
-      tbook = Spreadsheet.open(target)
+      if File.extname(target).eql?('.xlsx')
+        tbook = RubyXL::Parser.parse(File.expand_path(target))
+      else
+        tbook = Spreadsheet.open(target)
+      end
       sheet = tbook.worksheet(0)
       if new_column = cell(sheet.row(2), COLUMNS.size)
         raise "New column #{COLUMNS.size} (#{new_column})"
@@ -178,7 +181,11 @@ class SwissmedicDiff
       [known_regs, known_seqs, known_pacs, newest_rows]
     end
     def _known_data(latest, known_regs, known_seqs, known_pacs, newest_rows)
-      lbook = Spreadsheet.open(latest)
+      if File.extname(latest).eql?('.xlsx')
+        lbook = RubyXL::Parser.parse(File.expand_path(latest)).worksheets[0]
+      else
+        lbook = Spreadsheet.open(latest)
+      end
       idx, prr, prp = nil
       multiples = {}
       each_valid_row(lbook) { |row|
@@ -290,11 +297,18 @@ class SwissmedicDiff
     #return  ::
     def each_valid_row(spreadsheet)
       skipRows = rows_to_skip(spreadsheet)
-      worksheet = spreadsheet.worksheet(0)
-      worksheet.each(skipRows) {
+      if spreadsheet.class.eql?(RubyXL::Worksheet)
+        worksheet = spreadsheet
+      else
+        worksheet = spreadsheet.worksheet(0)
+      end
+      row_nr = 0
+      worksheet.each() {
         |row|
+        row_nr += 1
+        next if row_nr <= skipRows
         if row.size < COLUMNS.size/2 || row.select{|val| val==nil}.size > COLUMNS.size/2
-          $stdout.puts "Data missing in \n(line " + (row.idx+1).to_s + "): " + row.join(", ").to_s + "\n"
+          $stdout.puts "Data missing in \n(line " + (row_nr).to_s + "): " + row.join(", ").to_s + "\n"
           next
         end
         next if (cell(row, column(:production_science)) == 'Tierarzneimittel')
@@ -309,7 +323,11 @@ class SwissmedicDiff
       # Packungen.xls of swissmedic before October 2013 had  3 leading rows
       # Packungen.xls of swissmedic after  October 2013 have 4 leading rows
       j = 0
-      j += 1 while spreadsheet.worksheet(0).row(j)[0].to_i == 0
+      if spreadsheet.class.eql?(RubyXL::Worksheet)
+        j += 1 while spreadsheet[j][0] and spreadsheet[j][0].value.to_i == 0
+      else
+        j += 1 while spreadsheet.worksheet(0).row(j)[0].to_i == 0
+      end
       j
     end
     

data/lib/version.rb

@@ -0,0 +1,3 @@
+class SwissmedicDiff
+  VERSION = '0.1.6'
+end

data/swissmedic-diff.gemspec

@@ -1,6 +1,8 @@
+require File.expand_path('../lib/version', __FILE__)
+
 spec = Gem::Specification.new do |s|
    s.name        = "swissmedic-diff"
-   s.version     = "0.1.4"
+   s.version     = SwissmedicDiff::VERSION
    s.summary     = "Find out what Products have changed on the swiss healthcare market"
    s.description = "Compares two Excel Documents provided by Swissmedic and displays the salient differences"
    s.author      = "Hannes Wyss, Masaomi Hatakeyama"
@@ -13,6 +15,9 @@ spec = Gem::Specification.new do |s|
                    Dir.glob('test/data/*.xls')
    s.test_file   = "test/test_swissmedic-diff.rb"
    s.executables << 'swissmedic-diff'
+   s.add_dependency('rubyXL')
+   s.add_dependency('nokogiri')
+   s.add_dependency('rubyzip')
    s.add_dependency('spreadsheet')
    s.add_development_dependency "hoe"
    s.add_development_dependency "minitest"

data/test/test_swissmedic-diff.rb

@@ -6,6 +6,7 @@ $: << File.expand_path("../lib", File.dirname(__FILE__))
 
 require 'minitest/autorun'
 require 'swissmedic-diff'
+require 'pp'  
 
 module ODDB
   class SwissmedicPluginTest < Minitest::Test
@@ -25,27 +26,31 @@ module ODDB
                                 File.dirname(__FILE__)
       @workbook = Spreadsheet.open(@data)
     end
-
+    # This is not a unit test as it takes way too long (> 1 minute)
+    # Instead it might just tell you how to test with real data
+    def test_diff_xls_and_xlsx
+      @diff = SwissmedicDiff.new
+      last_month = File.expand_path 'data/Packungen.xls',  File.dirname(__FILE__)
+      this_month = File.expand_path 'data/Packungen-2014.xlsx',  File.dirname(__FILE__)
+      result = @diff.diff last_month, this_month, [:atc_class, :sequence_date]
+      assert(result.changes.flatten.index('Zulassungs-Nummer') == nil, "Should not find Zulassungs-Nummer in changes")
+      assert(result.news.first.index('00275'), "Should find 00275 in news")
+      assert(result.news.first.index('00277') == nil, "Should not find 00277 in news")
+      assert(result.news.flatten.index('Zulassungs-Nummer') == nil, "Should not find Zulassungs-Nummer in changes")
+    end
+    
     # This is not a unit test as it takes way too long (> 1 minute)
     # Instead it might just tell you how to test with real data
     def test_real_diff
-      require 'pp'  
       @diff = SwissmedicDiff.new
       last_month = File.expand_path 'data/Packungen-2013.08.16.xls',  File.dirname(__FILE__)
       this_month = File.expand_path 'data/Packungen-2013.11.04.xls',  File.dirname(__FILE__)
       result = @diff.diff last_month, this_month, [:atc_class, :sequence_date]
-      pp result.news.first
-      pp result.news.last
-      pp result.updates.first
-      pp result.replacements.first
-#      assert(result.changes.flatten.index('Zulassungs-Nummer') == nil, "Should not find Zulassungs-Nummer in changes")
-#      assert(result.news.flatten.index('Zulassungs-Nummer') == nil, "Should not find Zulassungs-Nummer in changes")
-#      assert(result.news.flatten.index('00277') == nil, "Should not find 00277 in news")
-      pp result.news.size
-      pp result.updates.size
-      pp result.replacements.size
+      assert(result.changes.flatten.index('Zulassungs-Nummer') == nil, "Should not find Zulassungs-Nummer in changes")
+      assert(result.news.flatten.index('Zulassungs-Nummer') == nil, "Should not find Zulassungs-Nummer in changes")
+      assert(result.news.flatten.index('00277') == nil, "Should not find 00277 in news")
     end if false
-    
+
     def test_iterate
       diff = SwissmedicDiff.new
       strings = []

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: swissmedic-diff
 version: !ruby/object:Gem::Version
-  version: 0.1.5
+  version: 0.1.6
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-12-11 00:00:00.000000000 Z
+date: 2014-01-14 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rdoc
@@ -76,9 +76,11 @@ files:
 - Rakefile
 - bin/swissmedic-diff
 - lib/swissmedic-diff.rb
+- lib/version.rb
 - setup.rb
 - swissmedic-diff.gemspec
 - test/data/Packungen-2013.10.14.xls
+- test/data/Packungen-2014.xlsx
 - test/data/Packungen.older.xls
 - test/data/Packungen.xls
 - test/data/Packungen_error_column.xls