ddi-parser 0.0.1 → 0.0.2

data/lib/ddi-parser.rb

@@ -5,7 +5,7 @@ require 'models/category'
 require 'models/category_statistic'
 require 'models/study'
 require 'models/study_date'
-require 'models/variable'
+require 'models/ddi_variable'
 require 'models/summary_stat'
 
 module DDI
@@ -18,7 +18,15 @@ module DDI
       catalog = DDI::Catalog.new
       study = DDI::Study.new
       study_info_hash = Hash.new
-      parser = LibXML::XML::Parser.file(ddi_file)
+      encode_type = `file --mime -br #{ddi_file}`.gsub(/\n/,"").split(';')[1].split('=')[1]
+      #have to convert to UTF-8 for libxml
+      contents = File.open(ddi_file).read
+      output = Iconv.conv("UTF-8", encode_type, contents)
+      converted_file = File.join(File.dirname(ddi_file), "converted_file.xml")
+      file = File.open(converted_file, 'w')
+      file.write(output)
+      file.close
+      parser = LibXML::XML::Parser.file(converted_file)
       doc = parser.parse
       studynodes = doc.find('//stdyDscr')
       abstracts = studynodes[0].find('//abstract')
@@ -36,15 +44,17 @@ module DDI
       date = studynodes[0].find('//sumDscr/collDate')
       date.each do |d|
         a = d.attributes
-        study_date = DDI::StudyDate.new
-        study_date.type = a.get_attribute('event').value.strip
-        study_date.date = a.get_attribute('date').value.strip
-        dates.push(study_date)
+        unless a.length == 0
+          study_date = DDI::StudyDate.new
+          study_date.type = a.get_attribute('event').value.strip
+          study_date.date = a.get_attribute('date').value.strip
+          dates.push(study_date)
+        end
       end
       study.dates = dates
-      study.sampling_procedure = studynodes[0].find('//sampProc')[0].first.content.strip unless studynodes[0].find('//sampProc')[0] == nil
+      study.sampling_procedure = studynodes[0].find('//sampProc')[0].first.content.strip unless studynodes[0].find('//sampProc')[0].children.size == 0
       # study.weight = studynodes[0].find('//sampProc')[0].first.content
-      study.variables = get_variable_information doc
+      study.ddi_variables = get_variable_information doc
       return study
     end
     
@@ -69,7 +79,7 @@ module DDI
       end
       vars = docnodes[0].find('//dataDscr/var')
       vars.each do |var|
-        variable = DDI::Variable.new
+        variable = DDI::DDIVariable.new
         var_attr = var.attributes
         variable.id = var_attr.get_attribute('ID').value.strip unless var_attr.get_attribute('ID') == nil
         variable.name = var_attr.get_attribute('name').value.strip unless var_attr.get_attribute('name') == nil
@@ -158,4 +168,4 @@ module DDI
     end
     
   end
-end
+end

data/lib/models/{variable.rb → ddi_variable.rb}

@@ -1,11 +1,11 @@
 module DDI
   
   #Information about a variable/column in a dataset
-  class Variable
+  class DDIVariable
   
     attr_reader :name, :label, :group, :id, :file, :interval, :max, :min, :question, :interview_instruction, :summary_stats, :categories
     attr_writer :name, :label, :group, :id, :file, :interval, :max, :min, :question, :interview_instruction, :summary_stats, :categories
       
   end
   
-end
+end

data/lib/models/study.rb

@@ -3,9 +3,9 @@ module DDI
   #Contains a set of variables and belongs to a catalog
   class Study
   
-    attr_reader :variables, :abstract, :title, :id, :dates, :sampling_procedure, :weight, :nesstar_id, :nesstar_uri
-    attr_writer :variables, :abstract, :title, :id, :dates, :sampling_procedure, :weight, :nesstar_id, :nesstar_uri
+    attr_reader :ddi_variables, :abstract, :title, :id, :dates, :sampling_procedure, :weight, :nesstar_id, :nesstar_uri
+    attr_writer :ddi_variables, :abstract, :title, :id, :dates, :sampling_procedure, :weight, :nesstar_id, :nesstar_uri
   
   end
   
-end
+end

data/lib/models/version.rb

@@ -1,5 +1,5 @@
 module DDI
   module Parser
-    VERSION = "0.0.1"
+    VERSION = "0.0.2"
   end
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: ddi-parser
 version: !ruby/object:Gem::Version 
-  hash: 29
-  prerelease: false
+  hash: 27
+  prerelease: 
   segments: 
   - 0
   - 0
-  - 1
-  version: 0.0.1
+  - 2
+  version: 0.0.2
 platform: ruby
 authors: 
 - Ian Dunlop
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-05-06 00:00:00 +01:00
+date: 2011-09-26 00:00:00 +01:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -45,14 +45,14 @@ extra_rdoc_files: []
 
 files: 
 - lib/ddi-parser.rb
+- lib/models/version.rb
+- lib/models/summary_stat.rb
+- lib/models/study.rb
 - lib/models/catalog.rb
 - lib/models/category.rb
 - lib/models/category_statistic.rb
-- lib/models/study.rb
+- lib/models/ddi_variable.rb
 - lib/models/study_date.rb
-- lib/models/summary_stat.rb
-- lib/models/variable.rb
-- lib/models/version.rb
 has_rdoc: true
 homepage: http://github.com/mygrid/ddi-parser
 licenses: []
@@ -83,7 +83,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
 requirements: []
 
 rubyforge_project: ddi-parser
-rubygems_version: 1.3.7
+rubygems_version: 1.6.2
 signing_key: 
 specification_version: 3
 summary: API for parsing ddi metadata files and returning results