davidrichards-etl 0.0.5 → 0.0.6

data/VERSION.yml

@@ -1,4 +1,4 @@
 --- 
 :major: 0
 :minor: 0
-:patch: 5
+:patch: 6

data/lib/etl/csv_et.rb

@@ -19,34 +19,8 @@ module CSV
       
       # Attempts to get a string from a file, a uri, or a string
       def extract
-        obj = self.options.fetch(:source, nil)
-        extract_locally(obj) or extract_remotely(obj) or extract_from_string(obj)
-        raise ArgumentError, "Could not determine what #{obj.inspect} was.  CSV::ET cannot work with this data." unless @raw
-      end
-      
-      # Handles local filename cases, reading the contents of the file.
-      def extract_locally(filename)
-        @raw = File.read(filename) if File.exist?(filename)
-        ET.logger.info "Extracted the data from from filesystem" if @raw
-        @raw ? true : false
-      end
-      
-      # Handles remote uri cases, reading the remote resource with open-uri, part of the Standard Library
-      def extract_remotely(uri)
-        begin
-          open(uri) {|f| @raw = f.read}
-          ET.logger.info "Extracted the data from a remote location."
-          return true
-        rescue
-          ET.logger.info "Tested whether #{uri} was a remote resource.  Failed to read it."
-          return false
-        end
-      end
-      
-      # If this is a string, assumes that the contents of the string are CSV contents.
-      def extract_from_string(string)
-        @raw = string if string.is_a?(String)
-        @raw ? true : false
+        source = self.options.fetch(:source, nil)
+        @raw = OpenContent::Extractor.process(source, ET.logger)
       end
 
       def transform

data/lib/etl/rdf_2_data_frame.rb

@@ -0,0 +1,19 @@
+# Requires data_frame (sudo gem install davidrichards-data_frame)
+# gem 'davidrichards-data_frame'
+# require 'data_frame'
+# 
+# # This is a simple tool that converts RDF to DataFrames.  It uses the
+# # subjects as the rows, the objects as the columns, and the predicates
+# # as the values.  This can make the data much more accessible by more
+# # analysis tools. 
+# class RDF2DataFrame < ETL
+#   
+#   def extract
+#     source = self.options.fetch(:source, nil)
+#     @raw = OpenContent::Extractor.process(source, RDF2DataFrame.logger)
+#   end
+#   
+#   def transform
+#     # TODO
+#   end
+# end

data/lib/helpers/open_content.rb

@@ -0,0 +1 @@
+Dir.glob("#{File.dirname(__FILE__)}/open_content/*.rb").each { |file| require file }

data/lib/helpers/open_content/extractor.rb

@@ -0,0 +1,43 @@
+require 'open-uri'
+module OpenContent
+  class Extractor
+    class << self
+      
+      attr_reader :logger
+      
+      def process(source, logger)
+        @logger = logger
+        extract_locally(source) or extract_remotely(source) or extract_from_string(source)
+        raise ArgumentError, "Could not determine what #{source.inspect} was.  Cannot extract this data." unless @raw
+        @raw
+      end
+      
+      protected
+        # Handles local filename cases, reading the contents of the file.
+        def extract_locally(filename)
+          @raw = File.read(filename) if File.exist?(filename)
+          self.logger.info "Extracted the data from from filesystem" if @raw
+          @raw ? true : false
+        end
+      
+        # Handles remote uri cases, reading the remote resource with open-uri, part of the Standard Library
+        def extract_remotely(uri)
+          begin
+            open(uri) {|f| @raw = f.read}
+            self.logger.info "Extracted the data from a remote location."
+            return true
+          rescue
+            self.logger.info "Tested whether #{uri} was a remote resource.  Failed to read it."
+            return false
+          end
+        end
+      
+        # If this is a string, assumes that the contents of the string are CSV contents.
+        def extract_from_string(string)
+          @raw = string if string.is_a?(String)
+          @raw ? true : false
+        end
+      
+    end
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: davidrichards-etl
 version: !ruby/object:Gem::Version 
-  version: 0.0.5
+  version: 0.0.6
 platform: ruby
 authors: 
 - David Richards
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2009-08-03 00:00:00 -07:00
+date: 2009-08-09 00:00:00 -07:00
 default_executable: etl
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -50,12 +50,16 @@ files:
 - lib/etl/bucket.rb
 - lib/etl/csv_et.rb
 - lib/etl/etl.rb
+- lib/etl/rdf_2_data_frame.rb
 - lib/etl/time_bucket.rb
 - lib/etl/xml_et.rb
 - lib/etl.rb
 - lib/helpers
 - lib/helpers/array.rb
 - lib/helpers/observation.rb
+- lib/helpers/open_content
+- lib/helpers/open_content/extractor.rb
+- lib/helpers/open_content.rb
 - lib/helpers/open_struct.rb
 - lib/helpers/string.rb
 - lib/helpers/symbol.rb