doc_wrapper 0.0.1 → 0.9.0

data/Gemfile.lock

@@ -0,0 +1,32 @@
+PATH
+  remote: .
+  specs:
+    doc_wrapper (0.0.1)
+      activesupport (>= 3.0.0)
+
+GEM
+  remote: http://rubygems.org/
+  specs:
+    ZenTest (4.4.2)
+    activesupport (3.0.0)
+    diff-lcs (1.1.2)
+    nokogiri (1.4.4)
+    rspec (2.5.0)
+      rspec-core (~> 2.5.0)
+      rspec-expectations (~> 2.5.0)
+      rspec-mocks (~> 2.5.0)
+    rspec-core (2.5.1)
+    rspec-expectations (2.5.0)
+      diff-lcs (~> 1.1.2)
+    rspec-mocks (2.5.0)
+
+PLATFORMS
+  ruby
+
+DEPENDENCIES
+  ZenTest
+  activesupport (>= 3.0.0)
+  bundler (>= 1.0.0)
+  doc_wrapper!
+  nokogiri
+  rspec (>= 2.0.0)

data/lib/doc_wrapper/base.rb

@@ -0,0 +1,7 @@
+module DocWrapper
+  module Base
+    def initialize (*args)
+      @documents = args.flatten
+    end
+  end
+end

data/lib/doc_wrapper/base_class_methods.rb

@@ -0,0 +1,71 @@
+module DocWrapper
+  module ClassMethods
+
+    # Add a property name to the singleton property_names attribute.
+    def add_property_name (property_name)
+      # Add the property name to the property_names collection.
+      self.property_names << property_name
+    end
+
+    # Add a property wrapper to the property_definitions Hash.
+    def add_property_wrapper (property_name, wrapper)
+      self.property_definitions[property_name] = wrapper
+    end
+
+    def add_property_accessor (property_name)
+      class_eval <<-END
+        def #{property_name.to_s}
+          property_definitions[:#{property_name.to_s}].property(documents)
+        end
+      END
+    end
+
+    # Set default options that all properties need.
+    def initialize_options (options)
+      # Make sure the options have a :document key with a value of 1.
+      # This forces all lookups to be for the 0th document in documents if
+      # the user did not specify an offset into the array.
+      { :document => 1 }.merge(options)
+    end
+
+    def add_property_definition (property_name, wrapper)
+      add_property_name(property_name)
+      add_property_wrapper(property_name, wrapper)
+      add_property_accessor(property_name)
+    end
+
+    # Create a typed property definition for a document wrapper. 
+    # The property_name must be a symbol.
+    def property (property_name, type, selector, options = {}, &block)
+      raise "Unhandled property type: #{type.to_s}" if ![:string, :date, :time, :boolean, :raw].include?(type)
+      add_property_definition(property_name,  build_property_definition(property_name, type, selector, initialize_options(options), block))
+    end
+    
+    def build_property_definition (property_name, type, selector, options, block)
+      DocWrapper.const_get("#{camelize(type.to_s)}PropertyDefinition").new(property_name, type, selector, initialize_options(options), block)
+    end
+
+    def camelize (string)
+      string.gsub(/\/(.?)/) { "::#{$1.upcase}" }.gsub(/(?:^|_)(.)/) { $1.upcase }
+    end
+    
+    def multi_property (property_name, selectors, options = {}, &block)
+      raise "Multi properties require a block" if block.nil?
+      add_property_definition(property_name, MultiPropertyDefinition.new(property_name, selectors, initialize_options(options), block))
+    end
+
+    def has_many (property_name, selector, klass, options = {})
+      options = initialize_options(options)
+      define_method property_name do
+        get_has_many( property_name, selector, klass, options)
+      end
+    end
+
+    def has_one (property_name, selector, klass, options = {})
+      options = initialize_options(options)
+      define_method(property_name) do
+        get_has_one(property_name, selector, klass, options)
+      end
+    end
+  end
+end

data/lib/doc_wrapper/base_property_definition.rb

@@ -0,0 +1,25 @@
+module DocWrapper
+  class BasePropertyDefinition
+    attr_accessor :property_name, :type, :selector, :options, :block
+
+    def initialize (property_name, type, selector, options, block)
+      @property_name = property_name
+      @type = type
+      @selector = selector
+      @options = options
+      @block = block
+    end
+
+    def property (documents)
+      begin
+        transform(documents[@options[:document] - 1].search(@selector))
+      rescue Nokogiri::CSS::SyntaxError
+        transform(documents[@options[:document] - 1].xpath(@selector))
+      end
+    end
+
+    def transform (result)
+      result
+    end
+  end
+end

data/lib/doc_wrapper/boolean_property_definition.rb

@@ -0,0 +1,13 @@
+module DocWrapper
+  class BooleanPropertyDefinition < InnerHtmlPropertyDefinition
+    def transform (result)
+      raise "BooleanPropertyDefinition: boolean_test is deprecated please use options[:parser] or a block." if options[:boolean_test]
+      if block
+        result = block.call(result)
+      else
+        result = result.blank? ? nil : (options[:parser] ? options[:parser].call(result) : options[:boolean_test].call(result) )
+      end
+      result
+    end
+  end
+end

data/lib/doc_wrapper/date_property_definition.rb

@@ -0,0 +1,12 @@
+module DocWrapper
+  class DatePropertyDefinition < InnerHtmlPropertyDefinition
+    def transform (result)
+      if block
+        result = block.call(result)
+      else
+        result = result.blank? ? nil : (options[:parser] ? options[:parser].call(result) : Date.parse(result))
+      end
+      result
+    end
+  end
+end

data/lib/doc_wrapper/inner_html_property_definition.rb

@@ -0,0 +1,11 @@
+module DocWrapper
+  class InnerHtmlPropertyDefinition < BasePropertyDefinition
+    def property (documents)
+      begin
+        transform(documents[@options[:document] - 1].search(@selector).inner_html.strip)
+      rescue Nokogiri::CSS::SyntaxError
+        transform(documents[@options[:document] - 1].xpath(@selector).inner_html.strip)
+      end
+    end
+  end
+end

data/lib/doc_wrapper/multi_property_definition.rb

@@ -0,0 +1,32 @@
+module DocWrapper
+  class MultiPropertyDefinition
+    attr_accessor :property_name, :selectors, :options, :block
+
+    def initialize (property_name, selectors, options, block)
+      @property_name = property_name
+      @selectors = selectors
+      @options = options
+      @block = block
+    end
+
+    def property (documents)
+      results = []
+      selectors.each do |selector|
+        nodes = documents[@options[:document] - 1].search(selector)
+        if nodes.respond_to? :inner_html
+          nodes.each do |node|
+            results << node.inner_html.strip
+          end
+        else
+          results << nodes.inner_html.strip
+        end
+      end
+      transform(results.flatten)
+    end
+
+    def transform (results)
+      return nil if results.size == 0
+      result = block.call(results)
+    end
+  end
+end

data/lib/doc_wrapper/properties.rb

@@ -0,0 +1,39 @@
+module DocWrapper
+  module Properties
+   
+    attr_reader :documents
+
+    def self.included (base)
+      base.extend(SattrAccessor)
+      base.extend(ClassMethods)
+      base.sattr_accessor :property_names
+      base.sattr_accessor :property_definitions
+      base.property_names = []
+      base.property_definitions = {}
+    end
+
+    def properties
+      result = Hash.new
+      property_names.each do |property|
+        begin
+          result[property] = send(property)
+        rescue StandardError => e
+          raise e
+        end
+      end
+      result
+    end
+
+    def get_has_many (property_name, selector, klass, options)
+      nodes = @documents.collect { |doc| result = doc.search(selector) ; result.blank? ? nil : result }.flatten.compact
+      start_row = options[:start_row] ? options[:start_row] : 0
+      end_row = options[:end_row] ? options[:end_row] : nodes.size - 1
+      nodes[start_row..end_row].collect { |node| klass.new(node) }
+    end
+
+    def get_has_one (property_name, selector, klass, options)
+      nodes = @documents.collect { |doc| result = doc.search(selector) ; result.blank? ? nil : result }.flatten.compact
+      klass.new(nodes)
+    end
+  end
+end

data/lib/doc_wrapper/raw_property_definition.rb

@@ -0,0 +1,8 @@
+module DocWrapper
+  class RawPropertyDefinition < BasePropertyDefinition
+    def transform (result)
+      result = block.call(result) if block
+      result
+    end
+  end
+end

data/lib/doc_wrapper/sattr_accessor.rb

@@ -0,0 +1,55 @@
+module DocWrapper
+
+  module SattrAccessor
+
+    def sattr_reader (sym, options = {})
+      options = { :inheritable => false }.merge(options)
+      class_eval(<<-END, __FILE__, __LINE__)
+        def self.#{sym}
+          @#{sym}
+        end
+
+        def #{sym}
+          result = self.class.#{sym}
+          #{
+            "
+              if result.is_a?(Array)
+                # Get the value from our ancestor if there is one.
+                if self.class.superclass.respond_to? :#{sym}
+                  result << self.class.superclass.#{sym}
+                end
+                result.flatten!
+              end
+              if result.is_a?(Hash)
+                # Get the value from our ancestor if there is one.
+                if self.class.superclass.respond_to? :#{sym}
+                  result = self.class.superclass.#{sym}.merge(result)
+                end
+              end
+            " if options[:inheritable] 
+          }
+          result
+        end
+      END
+    end
+
+    def sattr_writer (sym)
+      class_eval %Q{
+        def self.#{sym}= (value)
+          @#{sym} = value
+        end
+
+        def #{sym}= (value)
+          self.class.#{sym} = value
+        end
+      }
+    end
+
+    def sattr_accessor (sym, options = {})
+     sattr_reader(sym, options)
+     sattr_writer(sym)
+   end
+
+  end
+
+end

data/lib/doc_wrapper/string_property_definition.rb

@@ -0,0 +1,9 @@
+module DocWrapper
+  class StringPropertyDefinition < InnerHtmlPropertyDefinition
+    def transform (result)
+      result = options[:parser].call(result) if !result.blank? && options[:parser]
+      result = block.call(result) if block
+      result
+    end
+  end
+end

data/lib/doc_wrapper/time_property_definition.rb

@@ -0,0 +1,12 @@
+module DocWrapper
+  class TimePropertyDefinition < InnerHtmlPropertyDefinition
+    def transform (result)
+      if block
+        result = block.call(result)
+      else
+        result = result.blank? ? nil : (options[:parser] ? options[:parser].call(result) : Time.parse(result))
+      end
+      result
+    end
+  end
+end

data/lib/doc_wrapper/version.rb

@@ -1,3 +1,3 @@
 module DocWrapper
-  VERSION = "0.0.1"
+  VERSION = "0.9.0"
 end

data/spec/doc_wrapper_spec.rb

@@ -0,0 +1,171 @@
+require 'spec_helper'
+
+
+
+describe DocWrapper do
+
+  describe "The wrapped test documents" do
+    let(:doc1) do
+      Nokogiri::HTML(read_fixture_file("doc_wrapper_test.html"))
+    end
+    
+    let(:doc2) do
+      Nokogiri::HTML(read_fixture_file("doc_wrapper_test_2.html"))
+    end
+    
+    let(:document) do
+      TestDocWrapper.new([ doc1, doc2 ])
+    end
+    
+    describe "#properties" do
+      let(:document_properties) do
+        document.properties
+      end
+      
+      it { document_properties[:paragraph].should == 'A paragraph' }
+      it "should strip HTML   from string properties" do
+        pending
+        document_properties[:space_example].should == 'Space Example'
+      end
+      it { document_properties[:active].should be_true }
+      it { document_properties[:date].should == Date.parse('12-Dec-2009') }
+      it { document_properties[:date_with_block].should == 'result from block' }
+      it { document_properties[:time].should == Time.parse('12-Dec-2009 12:31 PM') }
+      it { document_properties[:parsed_date].should == Date.parse('12-Dec-2009') }
+      it { document_properties[:parsed_time].should == Time.parse('12-Dec-2009 12:31 PM') }
+      it { document_properties[:table_data].should == 'A table data' }
+      it { document_properties[:paragraph2].should == 'A second paragraph' }
+      it { document_properties[:combined_tds].should == 'A table data A table data' }
+      it { document_properties[:combined_date_and_time].should == Time.parse('12-Dec-2009 12:31 PM') }
+      it { document_properties[:arrayed_xpath].should == 'name1 position1 salary1' }
+      it { document_properties.size.should == 14 }
+    end
+
+    it { document.should respond_to(:paragraph) }
+    it { document.should respond_to(:table_data) }
+    it { document.should respond_to(:date) }
+    it { document.should respond_to(:parsed_date) }
+    it { document.should respond_to(:combined_tds) }
+    it { document.should respond_to(:combined_date_and_time) }
+    it { document.should respond_to(:arrayed_xpath) }
+
+    it { document.paragraph.should == 'A paragraph' }
+    it { document.paragraph2.should == 'A second paragraph' }
+    it { document.table_data.should == 'A table data' }
+    it { document.date.should == Date.parse('12-Dec-2009') }
+    it { document.parsed_date.should == Date.parse('12-Dec-2009') }
+    it { document.date_with_block.should == 'result from block' }
+    it { document.active.should == true }
+    it { document.time.should == Time.parse('12-Dec-2009 12:31 PM') }
+    it "should strip HTML   from string properties" do
+      pending
+      document.space_example.should == 'Space Example'
+    end
+    it { document.combined_tds.should == 'A table data A table data' }
+    it { document.combined_date_and_time.should == Time.parse('12-Dec-2009 12:31 PM') }
+    it { document.arrayed_xpath.should == 'name1 position1 salary1' }
+    it { document.raw_property.should == 'raw' }
+    it { document.line_items.size.should == 4 }
+    
+    describe "paragraph_wrapper" do
+      let(:paragraph_wrapper) do
+        document.paragraph_wrapper
+      end
+      
+      it { paragraph_wrapper.paragraph.should == 'A paragraph'}
+    end
+    
+    describe "line item 0" do
+      let(:line_items) do
+        document.line_items[0]
+      end
+      
+      it { line_items.name.should == 'name1' }
+      it { line_items.position.should == 'position1' }
+      it { line_items.salary.should == 'salary1' }
+    end
+    
+    describe "line item 1" do
+      let(:line_items) do
+        document.line_items[1]
+      end
+      
+      it { line_items.name.should == 'name2' }
+      it { line_items.position.should == 'position2' }
+      it { line_items.salary.should == 'salary2' }
+    end
+    
+    describe "has_one person" do
+      let(:person) do
+        document.person
+      end
+      
+      it { person.name.should == 'Mark Menard' }
+      it { person.home_town.should == 'Troy, NY' }
+    end
+    
+  end
+end
+
+class TestDocLineItem
+  include DocWrapper::Base
+  include DocWrapper::Properties
+
+  property :name, :string, "./td[1]"
+  property :position, :string, "./td[2]"
+  property :salary, :string, "./td[3]"
+
+end
+
+class PersonWrapper
+  include DocWrapper::Properties
+  include DocWrapper::Base
+  
+  property :name, :string, "./p[1]"
+  property :home_town, :string, "./p[2]"
+end
+
+class ParagraphWrapper
+  include DocWrapper::Properties
+  include DocWrapper::Base
+  
+  property :paragraph, :string, '.'
+end
+
+class TestDocWrapper
+  include DocWrapper::Properties
+  include DocWrapper::Base
+
+  property :paragraph, :string, "/html/body/p[1]"
+  has_one :paragraph_wrapper, "/html/body/p[1]", ParagraphWrapper
+  property :space_example, :string, "/html/body/p[5]" do |x|
+    x.strip
+  end
+  property :active, :boolean, "/html/body/p[2]", :parser => lambda { |x| x == 'Yes' }
+  property :date, :date, "/html/body/p[3]"
+  property :parsed_date, :date, "/html/body/p[3]", :parser => lambda { |x| Date.parse(x) }
+  property :date_with_block, :date, "/html/body/p[3]" do |x| 
+    "result from block"
+  end
+  property :time, :time, "/html/body/p[4]"
+  property :parsed_time, :time, "/html/body/p[4]", :parser => lambda { |x| Time.parse(x) }
+  property :table_data, :string, "/html/body/table[1]/tr/td[1]"
+  property :paragraph2, :string, "/html/body/p[1]", :document => 2
+  has_many :line_items, "/html/body/table[2]/tr", TestDocLineItem, :start_row => 1, :end_row => 4
+  multi_property :combined_tds, ["/html/body/table[1]/tr/td[1]", "/html/body/table[1]/tr/td[1]"] do |elements|
+    elements.join(" ")
+  end
+  multi_property :combined_date_and_time, ["/html/body/p[6]", "/html/body/p[7]"] do |elements|
+    Time.parse(elements.join(" "))
+  end
+  
+  # Example of a multi_property that uses an XPath that returns an array of elements.
+  multi_property :arrayed_xpath, ["/html/body/table[2]/tr[2]/td"] do |elements|
+    elements.join(" ")
+  end
+  property :raw_property, :raw, "/html/body/p[8]" do |ns|
+    ns[0].attribute("class").inner_html
+  end
+  
+  has_one :person, "/html/body/div", PersonWrapper
+end

data/spec/fixtures/doc_wrapper_test.html

@@ -0,0 +1,29 @@
+<html>
+  <body>
+    <p>A paragraph</p><!-- paragraph -->
+    <p>Yes</p><!-- active -->
+    <p>12-Dec-2009</p><!-- date -->
+    <p>12-Dec-2009 12:31 PM</p><!-- date and time -->
+    <p> &nbsp; Space Example&nbsp; &nbsp;</p>
+    <table>
+      <tr>
+        <td>A table data</td>
+      </tr>
+    </table>
+    <table class="has_many">
+      <tr><th colspan="3">header</th></tr>
+      <tr><td>name1</td><td>position1</td><td>salary1</td></tr>
+      <tr><td>name2</td><td>position2</td><td>salary2</td></tr>
+      <tr><td>name3</td><td>position3</td><td>salary3</td></tr>
+      <tr><td>name4</td><td>position4</td><td>salary4</td></tr>
+      <tr><th colspan="3">footer</th></tr>
+    </table>
+    <p>12-Dec-2009</p><!-- date -->
+    <p>12:31 PM</p><!-- time -->
+    <p class="raw">Raw paragraph.</p>
+		<div>
+			<p class="name">Mark Menard</p>
+			<p class="home_town">Troy, NY</p>
+		</div>
+  </body>
+</html>

data/spec/fixtures/doc_wrapper_test_2.html

@@ -0,0 +1,5 @@
+<html>
+  <body>
+    <p>A second paragraph</p>
+  </body>
+</html>

data/spec/spec_helper.rb

@@ -0,0 +1,6 @@
+require File.expand_path("../../lib/doc_wrapper", __FILE__)
+require 'nokogiri'
+
+def read_fixture_file (file_name)
+  File.open(File.join(File.dirname(__FILE__), "/fixtures/#{file_name}")).read
+end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: doc_wrapper
 version: !ruby/object:Gem::Version 
-  hash: 29
+  hash: 59
   prerelease: false
   segments: 
   - 0
+  - 9
   - 0
-  - 1
-  version: 0.0.1
+  version: 0.9.0
 platform: ruby
 authors: 
 - Mark Menard
@@ -106,10 +106,27 @@ extra_rdoc_files: []
 files: 
 - .gitignore
 - Gemfile
+- Gemfile.lock
 - Rakefile
 - doc_wrapper.gemspec
 - lib/doc_wrapper.rb
+- lib/doc_wrapper/base.rb
+- lib/doc_wrapper/base_class_methods.rb
+- lib/doc_wrapper/base_property_definition.rb
+- lib/doc_wrapper/boolean_property_definition.rb
+- lib/doc_wrapper/date_property_definition.rb
+- lib/doc_wrapper/inner_html_property_definition.rb
+- lib/doc_wrapper/multi_property_definition.rb
+- lib/doc_wrapper/properties.rb
+- lib/doc_wrapper/raw_property_definition.rb
+- lib/doc_wrapper/sattr_accessor.rb
+- lib/doc_wrapper/string_property_definition.rb
+- lib/doc_wrapper/time_property_definition.rb
 - lib/doc_wrapper/version.rb
+- spec/doc_wrapper_spec.rb
+- spec/fixtures/doc_wrapper_test.html
+- spec/fixtures/doc_wrapper_test_2.html
+- spec/spec_helper.rb
 has_rdoc: true
 homepage: http://rubygems.org/gems/doc_wrapper
 licenses: []