whos_got_dirt 0.0.2 → 0.0.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 98d4adb0f91ffac1ff9ceabe8232a3861e2f900a
-  data.tar.gz: 627276f14cba96da225df90590eb71267ef13c1a
+  metadata.gz: 063529b54e6a2c0acb0ab21f1eb44bce5683afe8
+  data.tar.gz: 428acf538ee358d701565695ca601078ba6db10b
 SHA512:
-  metadata.gz: d9e8cdc8e2076c071af133ac8abdcf97d0676c4edf6712c9591ab661ade771f5729abde4cef9766834fb49c282e6a6f1b134ab7daed572aaa55a9427b912184c
-  data.tar.gz: 1ff4a4d9c4afb312c5550afb83c48b503ecb82186fcb106240b6db7ca05187458d9fe067f1cef806e5c6f02841692150e0fb88ea21194e2560ed37f4b0010b89
+  metadata.gz: 79d13ce720b98a1e70f7adabab7ad6797faeb304edaca55400a40ed9c3b679376be61d1b303cb75bdc4b67538c20be89c5d6ae149c49dc055c9c2d1563e31d3f
+  data.tar.gz: 1d8433a1cf7d0b01784ddf2840baba3b93c3d515b0ce99e280eb74f54d25eee6d4db194cb230283f03a6bf3563c33590b520e4e072c1135458922b22e5feba85

data/Gemfile

@@ -2,6 +2,3 @@ source 'https://rubygems.org'
 
 # Specify your gem's dependencies in the gemspec
 gemspec
-
-# Remove once vcr 2.9.4 released.
-gem 'vcr', git: 'https://github.com/vcr/vcr.git'

data/README.md

@@ -1,4 +1,4 @@
-# Who's got dirt? A federated search API for entities and relations
+# Who's got dirt? A federated search API for influence data
 
 [![Gem Version](https://badge.fury.io/rb/whos_got_dirt.svg)](https://badge.fury.io/rb/whos_got_dirt)
 [![Build Status](https://secure.travis-ci.org/influencemapping/whos_got_dirt-gem.png)](https://travis-ci.org/influencemapping/whos_got_dirt-gem)

data/Rakefile

@@ -14,56 +14,3 @@ rescue LoadError
     abort 'YARD is not available. In order to run yard, you must: gem install yard'
   end
 end
-
-desc 'Fetch schemas, rewrite references, and store locally'
-task :schemas do
-  require 'json'
-  require 'open-uri'
-
-  require 'json-schema'
-
-  def process_value(value, definitions)
-    url = value['$ref']
-    if url
-      name = url.rpartition('/')[2].chomp('.json#')
-      value['$ref'] = "#/definitions/#{name}"
-      define(name, url, definitions)
-    end
-  end
-
-  def process_schema(url, definitions)
-    schema = JSON.load(open(url).read)
-    schema['properties'].each do |_,value|
-      process_value(value, definitions)
-      if value.key?('items')
-        process_value(value['items'], definitions)
-      end
-    end
-    schema
-  end
-
-  def define(name, url, definitions)
-    unless definitions.key?(name)
-      definitions[name] = {} # to avoid recursion
-      definitions[name] = process_schema(url, definitions)
-      definitions[name].delete('id')
-    end
-  end
-
-  definitions = {} # passed by reference
-
-  %w(organization person).each do |name|
-    define(name, "http://www.popoloproject.com/schemas/#{name}.json#", definitions)
-  end
-
-  schema = {
-    '$schema' => 'http://json-schema.org/draft-03/schema#',
-    'definitions' => definitions,
-  }
-
-  JSON::Validator.validate!(schema, {}, validate_schema: true)
-
-  File.open(File.expand_path(File.join('..', 'schemas', 'popolo.json'), __FILE__), 'w') do |f|
-    f.write(JSON.pretty_generate(schema))
-  end
-end

data/lib/whos_got_dirt/requests/entity/poderopedia.rb

@@ -33,10 +33,10 @@ module WhosGotDirt
           equal('user_key', 'poderopedia_api_key')
           equal('entity', 'type', valid: ['Person', 'Organization'], transform: lambda{|v|
             case v
-            when 'persona'
-              'Person'
-            when 'organizacion'
-              'Organization'
+            when 'Person'
+              'persona'
+            when 'Organization'
+              'organizacion'
             end
           })
 

data/lib/whos_got_dirt/response.rb

@@ -56,6 +56,13 @@ module WhosGotDirt
       raise NotImplementedError
     end
 
+    # Returns the error message in the response body.
+    #
+    # @return [String] the error message in the response body
+    def error_message
+      body
+    end
+
     # @!method initialize(response)
     #   Sets the response's response.
     #   @param [Faraday::Response] response a response
@@ -75,5 +82,9 @@ module WhosGotDirt
     # @!method status
     #   Returns the HTTP status code.
     #   @return [Fixnum] the HTTP status code
+
+    # @!method success?
+    #   Returns whether the request succeeded.
+    #   @return [Boolean] whether the request succeeded
   end
 end

data/lib/whos_got_dirt/responses/entity/little_sis.rb

@@ -26,7 +26,7 @@ module WhosGotDirt
             'url' => '/api_uri',
             'note' => 'LittleSis API detail',
           }],
-          'updated_at' => '/updated_at',
+          'updated_at' => date_formatter('updated_at', '/updated_at'),
 
           # Class-specific.
           'start_date' => lambda{|data|

data/lib/whos_got_dirt/responses/entity/open_corporates.rb

@@ -11,8 +11,6 @@ module WhosGotDirt
           'classification' => '/company_type',
           'founding_date' => '/incorporation_date',
           'dissolution_date' => '/dissolution_date',
-          'created_at' => '/created_at',
-          'updated_at' => '/updated_at',
           'other_names' => [{
             'name' => '/previous_names/company_name',
             'start_date' => '/previous_names/start_date',
@@ -37,14 +35,17 @@ module WhosGotDirt
           'sources' => [{
             'url' => '/sources/url',
             'note' => '/sources/publisher',
-            'retrieved_at' => '/sources/retrieved_at', # @todo check Dublin Core, etc.
+            # API-specific.
+            'retrieved_at' => '/sources/retrieved_at',
           }],
+          'created_at' => '/created_at',
+          'updated_at' => '/updated_at',
           # API-specific.
-          'branch_status' => '/branch_status', # @todo if boolean, use "branch" as in request
+          'branch_status' => '/branch_status',
           'current_status' => '/current_status',
           'inactive' => '/inactive',
           'jurisdiction_code' => '/jurisdiction_code',
-          'retrieved_at' => '/retrieved_at', # @todo check Dublin Core, etc.
+          'retrieved_at' => '/retrieved_at',
         }
 
         # Transforms the parsed response body into results.
@@ -52,7 +53,7 @@ module WhosGotDirt
         # @return [Array<Hash>] the results
         def to_a
           parsed_body['companies'].map do |data|
-            Result.new('Organization', renderer.result(data['company']), self).finalize!
+            Result.new('Entity', renderer.result(data['company']), self).finalize!
           end
         end
       end

data/lib/whos_got_dirt/responses/entity/open_duka.rb

@@ -36,6 +36,14 @@ module WhosGotDirt
             Result.new('Entity', renderer.result(data), self).finalize!
           end
         end
+
+        def success?
+          super && Array === parsed_body
+        end
+
+        def error_message
+          parsed_body['error']
+        end
       end
     end
   end

data/lib/whos_got_dirt/responses/entity/poderopedia.rb

@@ -8,11 +8,11 @@ module WhosGotDirt
         @template = {
           '@type' => 'Entity',
           'name' => '/alias',
+          'description' => '/shortBio',
           'identifiers' => [{
             'identifier' => '/id',
             'scheme' => 'Poderopedia',
           }],
-          'description' => '/shortBio'
         }
 
         # Parses the response body.

data/lib/whos_got_dirt/responses/helpers/little_sis.rb

@@ -6,6 +6,20 @@ module WhosGotDirt
           # @!attribute [r] count_field
           #   @return [Hash] the field storing the number of results
           attr_reader :count_field
+
+          # @private
+          def date_formatter(property, path)
+            return lambda{|data|
+              [property, JsonPointer.new(data, path).value.sub(' ', 'T') + 'Z']
+            }
+          end
+
+          # @private
+          def integer_formatter(property, path)
+            return lambda{|data|
+              [property, Integer(JsonPointer.new(data, path).value)]
+            }
+          end
         end
 
         # Parses the response body.
@@ -28,6 +42,11 @@ module WhosGotDirt
         def page
           Integer(parsed_body['Meta']['Parameters']['page'] || 1)
         end
+
+        # Returns the error message.
+        def error_message
+          parsed_body.strip
+        end
       end
     end
   end

data/lib/whos_got_dirt/responses/list/little_sis.rb

@@ -11,7 +11,7 @@ module WhosGotDirt
           '@type' => 'List',
           'name' => '/name',
           'description' => '/description',
-          'number_of_items' => '/num_entities',
+          'number_of_items' => integer_formatter('number_of_items', '/num_entities'),
           'item_list_order' => lambda{|data|
             v = JsonPointer.new(data, '/is_ranked').value
             if v == '1'
@@ -21,11 +21,11 @@ module WhosGotDirt
             end
             ['item_list_order', v]
           },
-          'updated_at' => '/updated_at',
           'identifiers' => [{
             'identifier' => '/id',
             'scheme' => 'LittleSis',
           }],
+          'updated_at' => date_formatter('updated_at', '/updated_at'),
         }
 
         # Transforms the parsed response body into results.

data/lib/whos_got_dirt/responses/list/open_corporates.rb

@@ -8,8 +8,6 @@ module WhosGotDirt
         @template = {
           '@type' => 'List',
           'name' => '/name',
-          'created_at' => '/created_at',
-          'updated_at' => '/updated_at',
           'identifiers' => [{
             'identifier' => '/wikipedia_id',
             'scheme' => 'Wikipedia',
@@ -18,6 +16,8 @@ module WhosGotDirt
             'url' => '/opencorporates_url',
             'note' => 'OpenCorporates page',
           }],
+          'created_at' => '/created_at',
+          'updated_at' => '/updated_at',
         }
 
         # Transforms the parsed response body into results.

data/lib/whos_got_dirt/responses/relation/open_oil.rb

@@ -5,12 +5,23 @@ module WhosGotDirt
       #
       # @see http://openoil.net/openoil-api/
       class OpenOil < Response
+        class << self
+          # @private
+          def date_formatter(property, path)
+            return lambda{|data|
+              v = JsonPointer.new(data, path).value
+              [property, v.empty? ? nil : v + 'T00:00:00Z']
+            }
+          end
+        end
+
         @template = {
           '@type' => 'Relation',
           'subject' => lambda{|data|
             v = JsonPointer.new(data, '/licensees').value
             ['subject', v.map{|licensee| {'name' => licensee}}]
           },
+          'name' => '/name',
           'identifiers' => [{
             'identifier' => '/identifier',
             'scheme' => 'OpenOil',
@@ -22,12 +33,11 @@ module WhosGotDirt
             'url' => '/url_wiki',
             'note' => 'OpenOil wiki page',
           }],
-          'name' => '/name',
-          'created_at' => '/source_date',
-          'updated_at' => '/retrieved_date',
           'sources' => [{
             'url' => '/source_document',
           }],
+          'created_at' => date_formatter('created_at', '/source_date'),
+          'updated_at' => date_formatter('updated_at', '/retrieved_date'),
           # API-specific.
           'additional_properties' => '/details',
           'country_code' => '/country',

data/lib/whos_got_dirt/result.rb

@@ -49,7 +49,7 @@ module WhosGotDirt
     def validate!(opts = {})
       # The code assumes that processing errors in reverse avoids re-indexing
       # issues when deleting items from arrays. If this assumption is invalid,
-      # we can delete items on a time and re-validate using this code skeleton:
+      # we can delete items one at a time and re-validate using this skeleton:
       #
       # begin
       #   Validator.validate(result, type)
@@ -82,7 +82,7 @@ module WhosGotDirt
           # the source having an integer instead of string value.
           pointer.value = pointer.value.to_s
         else
-          raise ValidationError.new(error.fetch(:message))
+          raise ValidationError.new("#{error.fetch(:message)} (#{pointer.value})")
         end
       end
 

data/lib/whos_got_dirt/validator.rb

@@ -34,7 +34,7 @@ module WhosGotDirt
       # Memoizes a validator using the schema.
       def validator
         # `JSON::Validator#initialize_schema` runs faster if given a `Hash`.
-        @@validator ||= JSON::Validator.new(JSON.load(File.read(File.expand_path(File.join('..', '..', '..', 'schemas', 'popolo.json'), __FILE__))), {}, {
+        @@validator ||= JSON::Validator.new(JSON.load(File.read(File.expand_path(File.join('..', '..', '..', 'schemas', 'schema.json'), __FILE__))), {}, {
           # Keep the cache - whatever it is.
           clear_cache: false,
           # It's safe to skip data parsing if the data is a `Hash`.

data/lib/whos_got_dirt/version.rb

@@ -1,3 +1,3 @@
 module WhosGotDirt
-  VERSION = "0.0.2"
+  VERSION = "0.0.3"
 end

data/schemas/schema.json

@@ -0,0 +1,388 @@
+{
+  "$schema": "http://json-schema.org/draft-03/schema#",
+  "definitions": {
+    "entity": {
+      "$schema": "http://json-schema.org/draft-03/schema#",
+      "type": "object",
+      "properties": {
+        "@type": {
+          "type": "string",
+          "enum": ["Entity"]
+        },
+        "type": {
+          "description": "An entity type, e.g. organization",
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "name": {
+          "description": "A primary name, e.g. a legally recognized name",
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "description": {
+          "description": "A description of the entity",
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "classification": {
+          "description": "An entity category, e.g. committee",
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "birth_date": {
+          "description": "A date of birth",
+          "type": [
+            "string",
+            "null"
+          ],
+          "pattern": "^[0-9]{4}(-[0-9]{2}){0,2}$"
+        },
+        "death_date": {
+          "description": "A date of death",
+          "type": [
+            "string",
+            "null"
+          ],
+          "pattern": "^[0-9]{4}(-[0-9]{2}){0,2}$"
+        },
+        "founding_date": {
+          "description": "A date of founding",
+          "type": [
+            "string",
+            "null"
+          ],
+          "pattern": "^[0-9]{4}(-[0-9]{2}){0,2}$"
+        },
+        "dissolution_date": {
+          "description": "A date of dissolution",
+          "type": [
+            "string",
+            "null"
+          ],
+          "pattern": "^[0-9]{4}(-[0-9]{2}){0,2}$"
+        },
+        "parent_id": {
+          "description": "The ID of the organization that contains this organization",
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "other_names": {
+          "description": "Alternate or former names",
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/other_name"
+          }
+        },
+        "identifiers": {
+          "description": "Issued identifiers",
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/identifier"
+          }
+        },
+        "contact_details": {
+          "description": "Means of contacting the entity",
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/contact_detail"
+          }
+        },
+        "links": {
+          "description": "URLs to documents about the entity",
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/link"
+          }
+        },
+        "sources": {
+          "description": "URLs to documents from which the resource is derived",
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/link"
+          }
+        },
+        "created_at": {
+          "description": "The time at which the resource was created",
+          "type": [
+            "string",
+            "null"
+          ],
+          "format": "date-time"
+        },
+        "updated_at": {
+          "description": "The time at which the resource was last modified",
+          "type": [
+            "string",
+            "null"
+          ],
+          "format": "date-time"
+        }
+      }
+    },
+    "relation": {
+      "$schema": "http://json-schema.org/draft-03/schema#",
+      "type": "object",
+      "properties": {
+        "@type": {
+          "type": "string",
+          "enum": ["Relation"]
+        },
+        "subject": {
+          "description": "The subject of the relation",
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/entity"
+          }
+        },
+        "object": {
+          "description": "The object of the relation",
+          "$ref": "#/definitions/entity"
+        },
+        "name": {
+          "description": "The name of the relation",
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "start_date": {
+          "description": "The date on which the relationship began",
+          "type": [
+            "string",
+            "null"
+          ],
+          "pattern": "^[0-9]{4}((-[0-9]{2}){0,2}|(-[0-9]{2}){2}(T[0-9]{2}(:[0-9]{2}(:[0-9]{2})?)?Z)?)$"
+        },
+        "end_date": {
+          "description": "The date on which the relationship ended",
+          "type": [
+            "string",
+            "null"
+          ],
+          "pattern": "^[0-9]{4}((-[0-9]{2}){0,2}|(-[0-9]{2}){2}(T[0-9]{2}(:[0-9]{2}(:[0-9]{2})?)?Z)?)$"
+        },
+        "identifiers": {
+          "description": "Issued identifiers",
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/identifier"
+          }
+        },
+        "links": {
+          "description": "URLs to documents about the list",
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/link"
+          }
+        },
+        "sources": {
+          "description": "URLs to documents from which the resource is derived",
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/link"
+          }
+        },
+        "created_at": {
+          "description": "The time at which the resource was created",
+          "type": [
+            "string",
+            "null"
+          ],
+          "format": "date-time"
+        },
+        "updated_at": {
+          "description": "The time at which the resource was last modified",
+          "type": [
+            "string",
+            "null"
+          ],
+          "format": "date-time"
+        }
+      }
+    },
+    "list": {
+      "$schema": "http://json-schema.org/draft-03/schema#",
+      "type": "object",
+      "properties": {
+        "@type": {
+          "type": "string",
+          "enum": ["List"]
+        },
+        "name": {
+          "description": "The name of the list",
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "description": {
+          "description": "A description of the list",
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "number_of_items": {
+          "description": "The number of items in the list",
+          "type": [
+            "integer",
+            "null"
+          ]
+        },
+        "item_list_order": {
+          "description": "The type of ordering (e.g. ascending, descending, unordered)",
+          "type": [
+            "string",
+            "null"
+          ],
+          "enum": [
+            "ascending",
+            "descending",
+            "unordered"
+          ]
+        },
+        "identifiers": {
+          "description": "Issued identifiers",
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/identifier"
+          }
+        },
+        "links": {
+          "description": "URLs to documents about the list",
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/link"
+          }
+        },
+        "sources": {
+          "description": "URLs to documents from which the resource is derived",
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/link"
+          }
+        },
+        "created_at": {
+          "description": "The time at which the resource was created",
+          "type": [
+            "string",
+            "null"
+          ],
+          "format": "date-time"
+        },
+        "updated_at": {
+          "description": "The time at which the resource was last modified",
+          "type": [
+            "string",
+            "null"
+          ],
+          "format": "date-time"
+        }
+      }
+    },
+    "other_name": {
+      "$schema": "http://json-schema.org/draft-03/schema#",
+      "title": "Other name",
+      "description": "An alternate or former name",
+      "type": "object",
+      "properties": {
+        "name": {
+          "description": "An alternate or former name",
+          "type": "string"
+        },
+        "start_date": {
+          "description": "The date on which the name was adopted",
+          "type": [
+            "string",
+            "null"
+          ],
+          "pattern": "^[0-9]{4}(-[0-9]{2}){0,2}$"
+        },
+        "end_date": {
+          "description": "The date on which the name was abandoned",
+          "type": [
+            "string",
+            "null"
+          ],
+          "pattern": "^[0-9]{4}(-[0-9]{2}){0,2}$"
+        },
+        "note": {
+          "description": "A note, e.g. 'Birth name'",
+          "type": [
+            "string",
+            "null"
+          ]
+        }
+      }
+    },
+    "identifier": {
+      "$schema": "http://json-schema.org/draft-03/schema#",
+      "title": "Identifier",
+      "description": "An issued identifier",
+      "type": "object",
+      "properties": {
+        "identifier": {
+          "description": "An issued identifier, e.g. a DUNS number",
+          "type": "string",
+          "required": true
+        },
+        "scheme": {
+          "description": "An identifier scheme, e.g. DUNS",
+          "type": [
+            "string",
+            "null"
+          ]
+        }
+      }
+    },
+    "contact_detail": {
+      "$schema": "http://json-schema.org/draft-03/schema#",
+      "title": "Contact detail",
+      "description": "A means of contacting an entity",
+      "type": "object",
+      "properties": {
+        "type": {
+          "description": "A type of medium, e.g. 'fax' or 'email'",
+          "type": "string",
+          "required": true
+        },
+        "value": {
+          "description": "A value, e.g. a phone number or email address",
+          "type": "string",
+          "required": true
+        }
+      }
+    },
+    "link": {
+      "$schema": "http://json-schema.org/draft-03/schema#",
+      "title": "Link",
+      "description": "A URL",
+      "type": "object",
+      "properties": {
+        "url": {
+          "description": "A URL",
+          "type": "string",
+          "format": "uri",
+          "required": true
+        },
+        "note": {
+          "description": "A note, e.g. 'Wikipedia page'",
+          "type": [
+            "string",
+            "null"
+          ]
+        }
+      }
+    }
+  }
+}