patinfo2csv 1.0.0

data/History.txt

@@ -0,0 +1,3 @@
+=== 1.0.0 / 22.03.2012
+
+* Initial commit for patinfo2csv

data/Manifest.txt

@@ -0,0 +1,9 @@
+History.txt
+Manifest.txt
+README.txt
+Rakefile
+bin/patinfo2csv
+lib/patinfo2csv.rb
+lib/patinfo2csv/cli.rb
+lib/patinfo2csv/converter.rb
+lib/version.rb

data/README.txt

@@ -0,0 +1,32 @@
+= Patinfo2csv
+
+* http://scm.ywesee.com/?p=patinfo2csv/.git;a=summary
+
+== Description
+
+Patinfo2csv extracts and converts patinfo.yaml to patinfo.csv;
+only 4 chapters(de, text) in patinfo.yaml are extracted.
+
+* effects
+* contra_indications
+* pregnancy
+* usage
+* composition
+
+You can specify the pharmacode, see `patinfo2csv --help`
+
+== Usage
+
+$ patinfo2csv --help
+Usage: patinfo2csv <input patinfo.yaml> <output patinfo.csv> <input pharmacode.txt>
+    -v, --version      Show version
+    -h, --help         Print this help message
+
+== Requirements
+
+* slop gem
+
+== Install
+
+* sudo gem install slop
+* sudo gem install patinfo2csv

data/Rakefile

@@ -0,0 +1,8 @@
+# -*- ruby -*-
+
+require 'rubygems'
+require 'hoe'
+
+Hoe.spec 'patinfo2csv' do
+  developer('Yasuhiro Asaka, Zeno R.R. Davatz', 'yasaka@ywesee.com, zdavatz@ywesee.com')
+end

data/bin/patinfo2csv

@@ -0,0 +1,45 @@
+#!/usr/bin/env ruby
+# encoding: utf-8
+
+require 'slop'
+require 'pathname'
+
+LANGUAGE = [
+  'de',
+  'fr'
+]
+
+argv = ARGV.dup
+opts = Slop.parse! argv, :help => true do
+  banner "Usage: patinfo2csv <input patinfo.yaml> <output patinfo.csv> <input pharmacode.txt> [-l (de|fr)]"
+  on :v, :version, 'Show version'
+  on :l, :lang,    'Language option (de|fr) default de', :argument => true
+end
+
+if opts[:lang]
+  lang = opts.options[:lang].argument_value.downcase
+  argv.delete(lang)
+else
+  lang = "de"
+end
+
+if argv.empty? \
+   or !argv.length.eql? 3   \
+   or !File.exist?(argv[0]) \
+   or !File.exist?(argv[2]) \
+   or !LANGUAGE.include?(lang)
+  puts opts.banner
+  exit
+end
+
+root = Pathname.new(__FILE__).realpath.parent.parent
+$:.unshift root.join('lib') if $0 == __FILE__
+
+if opts.version? then
+  require 'patinfo2csv/version'
+  puts "patinfo2csv " + Patinfo2csv::VERSION + "\n"
+  exit
+end
+
+require 'patinfo2csv'
+Patinfo2csv::CLI.run(argv[0], argv[2], argv[1], lang)

data/lib/patinfo2csv/cli.rb

@@ -0,0 +1,65 @@
+#!/usr/bin/env ruby
+# encoding: utf-8
+
+require 'yaml'
+require 'patinfo2csv'
+
+module Patinfo2csv
+  class CLI
+    class << self
+      # Param:: patinfo.yaml
+      # Param:: pharmacode.txt
+      # Param:: patinfo.csv
+      def run(patinfo, pharmacode, output, lang)
+        cv = Patinfo2csv::Converter.new()
+        cv.lang       = lang
+        cv.patinfo    = self.load_yaml(patinfo)
+        cv.pharmacode = self.parse_txt(pharmacode)
+        rows = cv.to_csv
+        self.output_rows(rows, output)
+        report(rows, cv.pharmacode, lang)
+      end
+      def load_yaml(patinfo)
+        # before yaml loading, replace escaped chars
+        file = ''
+        File.open(patinfo, "r:ascii:utf-8") { |f|
+          while line = f.gets
+            ESCAPED_STR_CODE_MAP.each do |code, char|
+              line.gsub!(code, char)
+            end
+            file << line
+          end
+        }
+        fh = StringIO.new file
+        YAML.load_documents(fh) # utf-8
+      end
+      def parse_txt(pharmacode)
+        codes = []
+        File.open(pharmacode, "r:utf-8") do |input|
+          while line=input.gets
+            codes << line.gsub(/[^\d]/, '').chomp
+          end
+        end
+        codes
+      end
+      def output_rows(rows, output)
+        return unless rows
+        open(output, "w:utf-8") do |out|
+          rows.each do |row|
+            out.print row, "\n"
+          end
+        end
+      end
+      def report(rows, nomatches, lang)
+        puts "patinfo[#{lang}] converted: #{(rows.length - 1 ).to_s} rows"
+        unless nomatches.empty?
+          puts "-----"
+          puts "There is no chapters of Patinfo of the following " \
+               "#{nomatches.length.to_s} codes"
+          puts "[#{nomatches.join(',')}]"
+          puts
+        end
+      end
+    end
+  end
+end

data/lib/patinfo2csv/converter.rb

@@ -0,0 +1,79 @@
+#!/usr/bin/env ruby
+# encoding: utf-8
+
+module Patinfo2csv
+  class Converter
+    attr_accessor :patinfo, :pharmacode, :lang
+    attr_reader :rows
+    CHAPTERS = [
+      :effects,
+      :contra_indications,
+      :pregnancy,
+      :usage,
+      :composition
+    ]
+    DELIMITER = ';'
+    def initialize
+      @lang       = "de"
+      @patinfo    = []
+      @pharmacode = []
+      @rows       = []
+    end
+    def to_csv
+      data = []
+      @patinfo.each_with_index do |row, index|
+        target = []
+        next if row['article_codes'].nil?
+        row['article_codes'].each do |code|
+          if @pharmacode.include?(code[:article_pcode])
+            target << code[:article_pcode]
+            @pharmacode.delete(code[:article_pcode])
+          end
+        end
+        next if target.empty?
+        chapters = extract_chapters(row)
+        next if chapters.empty?
+        target.each do |pcode|
+          @rows << [%Q!"#{pcode}"!, chapters].flatten.join(DELIMITER)
+        end
+      end
+      unless @rows.empty? #header
+        @rows.unshift([
+          "Pharmacode",
+          "Effects",
+          "Contra Indication",
+          "Pregnancy",
+          "Usage",
+          "Composition"
+        ].join(DELIMITER))
+      end
+      return @rows
+    end
+    def extract_chapters(row)
+      return nil if row['descriptions'][@lang].empty?
+      desc = row['descriptions'][@lang]
+      chapters = []
+      CHAPTERS.each do |chapter|
+        if (desc.has_key?(chapter.to_s) and !desc[chapter.to_s].nil?) and
+           !desc[chapter.to_s]['sections'].nil?
+          text = ''
+          desc[chapter.to_s]['sections'].each do |section|
+            text << section['subheading']
+            if section.has_key?('paragraphs')
+              section['paragraphs'].each do |paragraph|
+                unless paragraph['text'].empty?
+                  text << paragraph['text'].gsub(/\r\n|\r/, "\n") + "\n"
+                end
+              end
+            end
+            text << "\n"
+          end
+        else
+          text = "" #empty
+        end
+        chapters << %Q!"#{text.chomp}"!
+      end
+      chapters.join(DELIMITER)
+    end
+  end
+end

data/lib/patinfo2csv.rb

@@ -0,0 +1,91 @@
+#!/usr/bin/env ruby
+# encoding: utf-8
+
+require 'patinfo2csv/version'
+require 'patinfo2csv/cli'
+require 'patinfo2csv/converter'
+
+module Patinfo2csv
+  # FIXME
+  # escaped backslash and utf-8 code map
+  # for patinfo.yaml(ASII)
+  ESCAPED_STR_CODE_MAP = {
+    "\\x24"      => "$",
+    "\\x25"      => "%",
+    "\\x26"      => "&",
+    "\\x2D"      => "-",
+    "\\x5F"      => "_",
+    "\\xC2\\xA0" => " ",
+    "\\xC2\\xAB" => "«",
+    "\\xC2\\xBB" => "»",
+    "\\xC2\\xAE" => "®",
+    "\\xC2\\xA9" => "©",
+    "\\xC2\\xB5" => "µ",
+    "\\xC2\\xBD" => "½",
+    "\\xC2\\xBC" => "¼",
+    "\\xC2\\xBE" => "¾",
+    "\\xC3\\x82" => "Â",
+    "\\xC3\\x83" => "Ã", 
+    "\\xC3\\x84" => "Ä", 
+    "\\xC3\\x85" => "Å", 
+    "\\xC3\\x86" => "Æ", 
+    "\\xC3\\x87" => "Ç", 
+    "\\xC3\\x88" => "È", 
+    "\\xC3\\x89" => "É", 
+    "\\xC3\\x8A" => "Ê", 
+    "\\xC3\\x8B" => "Ë", 
+    "\\xC3\\x8C" => "Ì", 
+    "\\xC3\\x8D" => "Í", 
+    "\\xC3\\x8E" => "Î", 
+    "\\xC3\\x8F" => "Ï", 
+    "\\xC3\\x90" => "Ð",
+    "\\xC3\\x91" => "Ñ", 
+    "\\xC3\\x92" => "Ò", 
+    "\\xC3\\x93" => "Ó", 
+    "\\xC3\\x94" => "Ô", 
+    "\\xC3\\x95" => "Õ", 
+    "\\xC3\\x96" => "Ö", 
+    "\\xC3\\x97" => "×", 
+    "\\xC3\\x98" => "Ø", 
+    "\\xC3\\x99" => "Ù", 
+    "\\xC3\\x9A" => "Ú", 
+    "\\xC3\\x9B" => "Û", 
+    "\\xC3\\x9C" => "Ü", 
+    "\\xC3\\x9D" => "Ý", 
+    "\\xC3\\x9E" => "Þ", 
+    "\\xC3\\x9F" => "ß", 
+    "\\xC3\\xA0" => "à", 
+    "\\xC3\\xA1" => "á", 
+    "\\xC3\\xA2" => "â", 
+    "\\xC3\\xA3" => "ã", 
+    "\\xC3\\xA4" => "ä", 
+    "\\xC3\\xA5" => "å", 
+    "\\xC3\\xA6" => "æ", 
+    "\\xC3\\xA7" => "ç", 
+    "\\xC3\\xA8" => "è", 
+    "\\xC3\\xA9" => "é", 
+    "\\xC3\\xAA" => "ê", 
+    "\\xC3\\xAB" => "ë", 
+    "\\xC3\\xAC" => "ì", 
+    "\\xC3\\xAD" => "í", 
+    "\\xC3\\xAE" => "î", 
+    "\\xC3\\xAF" => "ï", 
+    "\\xC3\\xB0" => "ð", 
+    "\\xC3\\xB1" => "ñ", 
+    "\\xC3\\xB2" => "ò", 
+    "\\xC3\\xB3" => "ó", 
+    "\\xC3\\xB4" => "ô", 
+    "\\xC3\\xB5" => "õ", 
+    "\\xC3\\xB6" => "ö", 
+    "\\xC3\\xB7" => "÷", 
+    "\\xC3\\xB8" => "ø", 
+    "\\xC3\\xB9" => "ù", 
+    "\\xC3\\xBA" => "ú", 
+    "\\xC3\\xBB" => "û", 
+    "\\xC3\\xBC" => "ü", 
+    "\\xC3\\xBD" => "ý", 
+    "\\xC3\\xBE" => "þ", 
+    "\\xC3\\xBF" => "ÿ", 
+    "\\xE2\\x80\\x93" => "–",
+  }
+end

data/lib/version.rb

@@ -0,0 +1,3 @@
+module Patinfo2csv
+  VERSION = "1.0.0"
+end

metadata

@@ -0,0 +1,99 @@
+--- !ruby/object:Gem::Specification
+name: patinfo2csv
+version: !ruby/object:Gem::Version
+  version: 1.0.0
+  prerelease: 
+platform: ruby
+authors:
+- Yasuhiro Asaka, Zeno R.R. Davatz
+autorequire: 
+bindir: bin
+cert_chain: []
+date: 2012-03-22 00:00:00.000000000 Z
+dependencies:
+- !ruby/object:Gem::Dependency
+  name: rdoc
+  requirement: &25013040 !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ~>
+      - !ruby/object:Gem::Version
+        version: '3.10'
+  type: :development
+  prerelease: false
+  version_requirements: *25013040
+- !ruby/object:Gem::Dependency
+  name: hoe
+  requirement: &25012580 !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ~>
+      - !ruby/object:Gem::Version
+        version: '2.13'
+  type: :development
+  prerelease: false
+  version_requirements: *25012580
+description: ! 'Patinfo2csv extracts and converts patinfo.yaml to patinfo.csv;
+
+  only 4 chapters(de, text) in patinfo.yaml are extracted.
+
+
+  * effects
+
+  * contra_indications
+
+  * pregnancy
+
+  * usage
+
+  * composition
+
+
+  You can specify the pharmacode, see `patinfo2csv --help`'
+email:
+- yasaka@ywesee.com, zdavatz@ywesee.com
+executables:
+- patinfo2csv
+extensions: []
+extra_rdoc_files:
+- History.txt
+- Manifest.txt
+- README.txt
+files:
+- History.txt
+- Manifest.txt
+- README.txt
+- Rakefile
+- bin/patinfo2csv
+- lib/patinfo2csv.rb
+- lib/patinfo2csv/cli.rb
+- lib/patinfo2csv/converter.rb
+- lib/version.rb
+homepage: http://scm.ywesee.com/?p=patinfo2csv/.git;a=summary
+licenses: []
+post_install_message: 
+rdoc_options:
+- --main
+- README.txt
+require_paths:
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+required_rubygems_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+requirements: []
+rubyforge_project: patinfo2csv
+rubygems_version: 1.8.15
+signing_key: 
+specification_version: 3
+summary: Patinfo2csv extracts and converts patinfo.yaml to patinfo.csv; only 4 chapters(de,
+  text) in patinfo.yaml are extracted
+test_files: []