Raw page text

documentcloud.gemspec

@@ -4,7 +4,7 @@ $LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
 require 'document_cloud/version'
 
 Gem::Specification.new do |spec|
-  spec.add_dependency "rest-client", "~> 1.6.7"
+  spec.add_dependency "rest-client", "~> 1.8.0"
   spec.add_dependency 'multi_json', '~> 1.8'
 
   spec.name          = 'documentcloud'

lib/document_cloud.rb

@@ -1,9 +1,11 @@
 require 'rest_client'
 require 'multi_json'
 require 'date'
+require 'uri'
 
 require_relative 'document_cloud/document'
 require_relative 'document_cloud/project'
+require_relative 'document_cloud/page'
 require_relative 'document_cloud/search_results'
 require_relative 'document_cloud/default'
 require_relative 'document_cloud/configurable'
@@ -12,33 +14,33 @@
 module DocumentCloud
   class << self
     include DocumentCloud::Configurable
-    
+
     # Delegate to a DocumentCloud::Client
     #
     # @return [DocumentCloud::Client]
     def client
       @client = DocumentCloud::Client.new(credentials) unless defined?(@client)
       @client
     end
-    
+
     # Has a client been initialized on the DocumentCloud module?
     #
     # @return [Boolean]
     def client?
       !!@client
     end
-    
-    
+
+
     def respond_to?(method_name, include_private=false)
       client.respond_to?(method_name, include_private) || super
     end
-    
+
     private
 
-      def method_missing(method_name, *args, &block)
-        return super unless client.respond_to?(method_name)
-        client.send(method_name, *args, &block)
-      end
-    
+    def method_missing(method_name, *args, &block)
+      return super unless client.respond_to?(method_name)
+      client.send(method_name, *args, &block)
+    end
+
   end
-end
+end

lib/document_cloud/api/page.rb

@@ -0,0 +1,26 @@
+module DocumentCloud
+  module API
+    module Page
+      include DocumentCloud::API::Utils
+
+      # Fetch document entities
+      #
+      # @param id [String] The document id
+      # @returns [Hash] The fetched document entities
+      def page(id, page_number, url)
+        url = URI.escape(replace_page_number(url, page_number))
+        text = get(url, {}, false)
+        DocumentCloud::Page.new(id: id,
+                                page_number: page_number,
+                                text: text)
+      end
+
+      private
+
+      def replace_page_number(url, page_number)
+        url.sub(/\{page\}/, page_number.to_s)
+      end
+
+    end
+  end
+end

lib/document_cloud/client.rb

@@ -3,6 +3,7 @@
 require_relative 'api/search'
 require_relative 'api/upload'
 require_relative 'api/document'
+require_relative 'api/page'
 require_relative 'api/update'
 require_relative 'api/destroy'
 require_relative 'api/entities'
@@ -20,41 +21,58 @@ class Client
     include DocumentCloud::API::Destroy
     include DocumentCloud::API::Entities
     include DocumentCloud::API::Projects
+    include DocumentCloud::API::Page
     include DocumentCloud::API::CreateProject
     include DocumentCloud::API::UpdateProject
     include DocumentCloud::Configurable
-    
+
     def initialize(options={})
       DocumentCloud::Configurable.keys.each do |key|
         instance_variable_set(:"@#{key}", options[key] || DocumentCloud.instance_variable_get(:"@#{key}"))
       end
     end
-    
+
     # Perform HTTP GET request
-    def get(path, params={})
-      RestClient.get request_base+path, {params: params}
+    def get(path, params = {}, use_request_base = true)
+      url = construct_url(path, use_request_base)
+      resource = construct_resource(url)
+      resource.get params
     end
-    
+
     # Perform HTTP POST request
-    def post(path, params={})
-      RestClient.post request_base+path, params
+    def post(path, params = {}, use_request_base = true)
+      url = construct_url(path, use_request_base)
+      resource = construct_resource(url)
+      resource.post params
     end
-    
+
     # Perform HTTP PUT request
-    def put(path, params={})
-      RestClient.put request_base+path, params
+    def put(path, params = {}, use_request_base = true)
+      url = construct_url(path, use_request_base)
+      resource = construct_resource(url)
+      resource.put params
     end
-    
+
     # Perform HTTP DELETE request
-    def delete(path)
-      RestClient.delete request_base+path
+    def delete(params = {}, use_request_base = true)
+      url = construct_url(path, use_request_base)
+      resource = construct_resource(url)
+      resource.delete
     end
-    
+
     private
-
-      def request_base
-        "#{DocumentCloud::Default.http_mode}://#{@email}:#{@password}@#{DocumentCloud::Default.endpoint}"
-      end
-
+
+    def construct_url(path, use_request_base)
+      use_request_base ? request_base + path : path
+    end
+
+    def construct_resource(url)
+      RestClient::Resource.new(url, user: @email, password: @password)
+    end
+
+    def request_base
+      "#{DocumentCloud::Default.http_mode}://#{DocumentCloud::Default.endpoint}"
+    end
+
   end
-end
+end

lib/document_cloud/configurable.rb

@@ -1,39 +1,27 @@
-require 'cgi'
 module DocumentCloud
   module Configurable
     attr_writer :email, :password
-    
+
     class << self
       def keys
         @keys ||= [:email, :password]
       end
     end
-    
+
     # Allow block configuration
     def configure
       yield self
-      format_email!
-      format_password!
       self
     end
-    
+
     private
-
-      # @return [Hash]
-      def credentials
-        {
-          email:    @email,
-          password: @password 
-        }
-      end
-
-      # Ensure email is correct format for RestClient posts
-      def format_email!
-        @email = CGI.escape @email
-      end
-      def format_password!
-        @password = CGI.escape @password
-      end
-
+
+    # @return [Hash]
+    def credentials
+      {
+        email:    @email,
+        password: @password
+      }
+    end
   end
-end
+end

lib/document_cloud/document.rb

@@ -2,7 +2,7 @@ module DocumentCloud
   class Document
     attr_reader :id, :title, :access, :pages, :description, :source,
       :canonical_url, :language, :display_language, :created_at, :updated_at
-    
+
     def initialize(attrs={})
       @id               = attrs[:id]
       @title            = attrs[:title]
@@ -17,35 +17,48 @@ def initialize(attrs={})
       @updated_at       = DateTime.parse(attrs[:updated_at])
       @resources        = attrs[:resources]
     end
-    
+
     def pdf
       @resources[:pdf]
     end
-    
+
     def print_annotations
       @resources[:print_annotations]
     end
-    
+
     def related_article
       @resources[:related_article]
     end
-    
+
     def text
       @resources[:text]
     end
-
+
+    def page(page_number)
+      return unless valid_page_number?(page_number)
+      @page ||= {}
+      @page[page_number] ||= DocumentCloud.page(@id, page_number, @resources[:page][:text])
+    end
+
     def thumbnail
       @resources[:thumbnail]
     end
-    
+
     def image(page, size=1)
       @resources[:page][:image].gsub(/\{page\}/, page.to_s).gsub(/\{size\}/,size.to_s)
     end
-    
+
     def entities
       @entities ||= DocumentCloud.entities(@id)
-      @entities
     end
-
+
+    private
+
+    def valid_page_number?(page_number)
+      page_number.is_a? Integer and
+        page_number <= pages and
+        page_number > 0
+    end
+
   end
-end
+end

lib/document_cloud/page.rb

@@ -0,0 +1,11 @@
+module DocumentCloud
+  class Page
+    attr_reader :document_id, :page_number, :text
+
+    def initialize(attrs={})
+      @document_id      = attrs[:id]
+      @page_number      = attrs[:page_number]
+      @text             = attrs[:text]
+    end
+  end
+end