turbot-runner 0.2.31 → 0.2.35

data/schema/schemas/licence-schema.json

@@ -107,4 +107,4 @@
     "licence_issuer",
     "jurisdiction_of_licence"
   ]
-}
+}

data/schema/schemas/primary-data-schema.json

@@ -17,4 +17,4 @@
     "sample_date"
   ],
   "additionalProperties": true
-}
+}

data/schema/schemas/register-entry-schema.json

@@ -5,7 +5,9 @@
     "subject_entity": {
       "$ref": "includes/entity.json"
     },
-    "register": {"$ref": "#/definitions/register"},
+    "register": {
+      "$ref": "#/definitions/register"
+    },
     "identifier": {
       "description": "A unique identifier used by the register to identifier the register entry. In some cases – e.g. Charity Register for England & Wales, these identifiers (in this case known as the 'charity number') are used outside of the context of the register",
       "type": "string"
@@ -110,6 +112,5 @@
         "title"
       ]
     }
-
   }
-}
+}

data/schema/schemas/sanctioned-entity-schema.json

@@ -13,8 +13,9 @@
         "DebarredSupplier"
       ]
     },
-    "sanctions_list": {"$ref": "#/definitions/sanctions_list"},
-
+    "sanctions_list": {
+      "$ref": "#/definitions/sanctions_list"
+    },
     "sanctions": {
       "type": "array",
       "items": {
@@ -109,7 +110,5 @@
         "title"
       ]
     }
-
   }
-
-}
+}

data/schema/schemas/share-parcel-schema.json

@@ -19,4 +19,4 @@
       }
     }
   ]
-}
+}

data/schema/schemas/simple-financial-payment-schema.json

@@ -124,4 +124,4 @@
     "date",
     "currency"
   ]
-}
+}

data/schema/schemas/simple-licence-schema.json

@@ -83,4 +83,4 @@
     "company_name",
     "company_jurisdiction"
   ]
-}
+}

data/schema/schemas/simple-subsidiary-schema.json

@@ -86,4 +86,4 @@
     "subsidiary_name",
     "subsidiary_jurisdiction"
   ]
-}
+}

data/schema/schemas/subsequent-registration-schema.json

@@ -0,0 +1,89 @@
+{
+  "$schema": "http://json-schema.org/draft-04/schema#",
+  "description": "A representation of entity registration changing over time",
+  "type": "object",
+  "properties": {
+    "data_type": {
+      "enum": [
+        "subsequent-registration"
+      ]
+    },
+    "previous_entity": {
+      "description": "the previous entity",
+      "$ref": "includes/entity.json"
+    },
+    "subsequent_entity": {
+      "description": "the subsequent entity",
+      "$ref": "includes/entity.json"
+    },
+    "subsequent_registration_start_date": {
+      "description": "date when subsequent registration started",
+      "type": "string",
+      "format": "date"
+    },
+    "previous_registration_end_date": {
+      "description": "date when previous registration ended",
+      "type": "string",
+      "format": "date"
+    },
+    "publication_date": {
+      "description": "the publication date of the filing/notice that gives details of the alternate registration",
+      "type": "string",
+      "format": "date"
+    },
+    "start_date": {
+      "description": "date when subsequent registration was valid from",
+      "type": "string",
+      "format": "date"
+    },
+    "start_date_type": {
+      "enum": [
+        "at",
+        "before",
+        "after"
+      ]
+    },
+    "sample_date": {
+      "description": "date when subsequent registration was sampled",
+      "type": "string",
+      "format": "date"
+    },
+    "retrieved_at": {
+      "description": "date when subsequent registration was retrieved",
+      "type": "string",
+      "format": "date"
+    },
+    "source_url": {
+      "description": "URL of the source of the data (e.g. download URL), or if there is not persistent URL the page from which it can be found (e.g. search page)",
+      "type": "string"
+    },
+    "confidence": {
+      "description": "Confidence in accuracy of data",
+      "enum": [
+        "HIGH",
+        "MEDIUM",
+        "LOW"
+      ]
+    }
+  },
+  "additionalProperties": false,
+  "required": [
+    "data_type",
+    "previous_entity",
+    "subsequent_entity",
+    "retrieved_at",
+    "confidence"
+  ],
+  "anyOf": [
+    {
+      "required": [
+        "start_date"
+      ]
+    },
+    {
+      "required": [
+        "sample_date"
+      ]
+    }
+  ]
+}

data/schema/schemas/subsidiary-relationship-schema.json

@@ -47,4 +47,4 @@
     "data",
     "sample_date"
   ]
-}
+}

data/schema/schemas/supplier-relationship-schema.json

@@ -0,0 +1,82 @@
+{
+  "$schema": "http://json-schema.org/draft-04/schema#",
+  "type": "object",
+  "properties": {
+    "data_type": {
+      "enum": [
+        "supplier-relationship"
+      ]
+    },
+    "customer": {
+      "$ref": "includes/entity.json"
+    },
+    "supplier": {
+      "$ref": "includes/entity.json"
+    },
+    "start_date": {
+      "type": "string",
+      "format": "date"
+    },
+    "start_date_type": {
+      "enum": [
+        "at",
+        "before",
+        "after"
+      ]
+    },
+    "end_date": {
+      "type": "string",
+      "format": "date"
+    },
+    "end_date_type": {
+      "enum": [
+        "at",
+        "before",
+        "after"
+      ]
+    },
+    "sample_date": {
+      "type": "string",
+      "format": "date"
+    },
+    "retrieved_at": {
+      "type": "string",
+      "format": "date"
+    },
+    "source_url": {
+      "type": "string"
+    },
+    "confidence": {
+      "enum": [
+        "HIGH",
+        "MEDIUM",
+        "LOW"
+      ]
+    }
+  },
+  "additionalProperties": false,
+  "required": [
+    "data_type",
+    "supplier",
+    "customer",
+    "retrieved_at",
+    "confidence"
+  ],
+  "anyOf": [
+    {
+      "required": [
+        "start_date"
+      ]
+    },
+    {
+      "required": [
+        "sample_date"
+      ]
+    },
+    {
+      "required": [
+        "end_date"
+      ]
+    }
+  ]
+}

data/schema/schemas/trademark-registration-schema.json

@@ -154,7 +154,9 @@
       "properties": {
         "code_scheme_id": {
           "description": "The scheme of the classification",
-          "enum": ["wipo_nice"]
+          "enum": [
+            "wipo_nice"
+          ]
         },
         "code": {
           "description": "The classification code",

data/spec/lib/runner_spec.rb

@@ -7,6 +7,10 @@ describe TurbotRunner::Runner do
     puts 'If all specs passed, you should now run `ruby spec/manual.rb`'
   end
 
+  after do
+    FileUtils.rm_rf(File.join(@runner.base_directory, "output")) if @runner
+  end
+
   describe '#run' do
     context 'with a bot written in ruby' do
       before do
@@ -69,11 +73,14 @@ describe TurbotRunner::Runner do
 
     context 'with a bot that logs' do
       context 'when logging to file enabled' do
+        before do
+          @runner = test_runner('logging-bot', :log_to_file => true)
+        end
+
         it 'logs to file' do
           expected_log = "doing...\ndone\n"
-          runner = test_runner('logging-bot', :log_to_file => true)
-          runner.run
-          expect(runner).to have_error_output_matching('scraper', expected_log)
+          @runner.run
+          expect(@runner).to have_error_output_matching('scraper', expected_log)
         end
       end
 
@@ -201,15 +208,21 @@ describe TurbotRunner::Runner do
     end
 
     context 'with a scraper that produces an invalid record' do
-      it 'returns false' do
+      before do
         @runner = test_runner('invalid-record-bot')
+      end
+
+      it 'returns false' do
         expect(@runner).to fail_in_scraper
       end
     end
 
     context 'with a scraper that produces invalid JSON' do
-      it 'returns false' do
+      before do
         @runner = test_runner('invalid-json-bot')
+      end
+
+      it 'returns false' do
         expect(@runner).to fail_in_scraper
       end
     end
@@ -220,11 +233,14 @@ describe TurbotRunner::Runner do
       # output file is created; however, the way we're redirecting
       # stdout using the shell means the file doesn't get created
       # until
-      it 'returns false' do
+      before do
         @runner = test_runner('bot-with-pause',
           :timeout => 1,
           :log_to_file => true
         )
+      end
+
+      it 'returns false' do
         expect(@runner).to fail_in_scraper
       end
     end
@@ -315,51 +331,67 @@ describe TurbotRunner::Runner do
       @handler = Handler.new
     end
 
-    it 'calls handler once for each line of output' do
-      test_runner('bot-with-transformer').run
+    context 'with a bot that runs correctly' do
+      before do
+        @runner = test_runner('bot-with-transformer')
+        @runner.run
+      end
 
-      runner = test_runner('bot-with-transformer',
-        :record_handler => @handler
-      )
+      it 'calls handler once for each line of output' do
+        runner = test_runner('bot-with-transformer',
+          :record_handler => @handler
+        )
 
-      runner.process_output
-      expect(@handler.records_seen['primary data']).to eq(10)
-      expect(@handler.records_seen['simple-licence']).to eq(10)
-    end
+        runner.process_output
+        expect(@handler.records_seen['primary data']).to eq(10)
+        expect(@handler.records_seen['simple-licence']).to eq(10)
+      end
 
-    it 'passes opts to processor.process' do
-      test_runner('bot-with-transformer').run
-      runner = test_runner('bot-with-transformer',
-        :record_handler => @handler
-      )
-      opts = {frob: 5}
-      processor = double('processor')
-      allow(TurbotRunner::Processor).to receive(:new).and_return(processor)
-      expect(processor).to receive(:process).with(anything, opts).at_least(:once)
-      runner.process_output(opts)
+      it 'passes opts to processor.process' do
+        runner = test_runner('bot-with-transformer',
+          :record_handler => @handler
+        )
+        opts = {frob: 5}
+        processor = double('processor')
+        allow(TurbotRunner::Processor).to receive(:new).and_return(processor)
+        expect(processor).to receive(:process).with(anything, opts).at_least(:once)
+        runner.process_output(opts)
+      end
+
+      context 'when skip_data_types is set' do
+        it 'skips the data type' do
+          runner = test_runner('bot-with-transformer',
+            :record_handler => @handler
+          )
+
+          runner.process_output(skip_data_types: ['primary data'])
+          expect(@handler.records_seen['primary data']).to eq(0)
+          expect(@handler.records_seen['simple-licence']).to eq(10)
+        end
+      end
     end
 
-    it 'can cope when scraper has failed immediately' do
-      test_runner('bot-that-crashes-immediately').run
+    context 'with a bot that crashes immediately' do
+      before do
+        @runner = test_runner('bot-that-crashes-immediately')
+        @runner.run
+      end
 
-      runner = test_runner('bot-that-crashes-immediately',
-        :record_handler => @handler
-      )
+      it 'can cope with the empty files' do
+        runner = test_runner('bot-that-crashes-immediately',
+          :record_handler => @handler
+        )
 
-      runner.process_output
+        runner.process_output
+      end
     end
 
-    context 'when skip_data_types is set' do
-      it 'skips the data type' do
-        test_runner('bot-with-transformer').run
-
+    context 'when no bot has run' do
+      it 'proceeds without errors' do
         runner = test_runner('bot-with-transformer',
           :record_handler => @handler
         )
-
-        runner.process_output(skip_data_types: ['primary data'])
-        expect(@handler.records_seen['primary data']).to eq(0)
-        expect(@handler.records_seen['simple-licence']).to eq(10)
+        runner.process_output
       end
     end
   end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: turbot-runner
 version: !ruby/object:Gem::Version
-  version: 0.2.31
+  version: 0.2.35
 platform: ruby
 authors:
 - OpenCorporates
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2017-12-15 00:00:00.000000000 Z
+date: 2021-09-08 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -68,7 +68,9 @@ dependencies:
         version: 3.4.0
 description: 
 email: bots@opencorporates.com
-executables: []
+executables:
+- bundle
+- rspec
 extensions: []
 extra_rdoc_files: []
 files:
@@ -81,6 +83,8 @@ files:
 - README.md
 - Rakefile
 - appveyor.yml
+- bin/bundle
+- bin/rspec
 - lib/turbot_runner.rb
 - lib/turbot_runner/base_handler.rb
 - lib/turbot_runner/exceptions.rb
@@ -92,41 +96,43 @@ files:
 - lib/turbot_runner/validator.rb
 - lib/turbot_runner/version.rb
 - schema/schemas/accounts-statement-schema.json
+- schema/schemas/alternate-registration-schema.json
+- schema/schemas/alternative-name-schema.json
 - schema/schemas/company-schema.json
 - schema/schemas/control-statement-schema.json
 - schema/schemas/filing-schema.json
 - schema/schemas/financial-payment-schema.json
 - schema/schemas/gazette-notice-schema.json
-- schema/schemas/includes/accounts_element.json
+- schema/schemas/includes/accounts-element.json
 - schema/schemas/includes/address-with-type.json
 - schema/schemas/includes/address.json
-- schema/schemas/includes/alternative_name.json
+- schema/schemas/includes/alternative-name.json
 - schema/schemas/includes/base-statement.json
 - schema/schemas/includes/classification.json
 - schema/schemas/includes/company-for-nesting.json
 - schema/schemas/includes/company.json
 - schema/schemas/includes/date.json
 - schema/schemas/includes/entity.json
+- schema/schemas/includes/filing-document.json
 - schema/schemas/includes/filing.json
-- schema/schemas/includes/filing_document.json
 - schema/schemas/includes/financial-payment-data-object.json
 - schema/schemas/includes/identifier.json
-- schema/schemas/includes/industry_code.json
+- schema/schemas/includes/industry-code.json
 - schema/schemas/includes/legislation.json
 - schema/schemas/includes/licence-data-object.json
 - schema/schemas/includes/officer.json
 - schema/schemas/includes/organisation.json
 - schema/schemas/includes/permission.json
+- schema/schemas/includes/person-name.json
 - schema/schemas/includes/person.json
-- schema/schemas/includes/person_name.json
-- schema/schemas/includes/previous_name.json
+- schema/schemas/includes/previous-name.json
 - schema/schemas/includes/range.json
 - schema/schemas/includes/sanction.json
 - schema/schemas/includes/share-parcel-data.json
 - schema/schemas/includes/share-parcel.json
 - schema/schemas/includes/subsidiary-relationship-data.json
 - schema/schemas/includes/total-shares.json
-- schema/schemas/includes/unknown_entity_type.json
+- schema/schemas/includes/unknown-entity-type.json
 - schema/schemas/licence-schema.json
 - schema/schemas/primary-data-schema.json
 - schema/schemas/register-entry-schema.json
@@ -135,7 +141,9 @@ files:
 - schema/schemas/simple-financial-payment-schema.json
 - schema/schemas/simple-licence-schema.json
 - schema/schemas/simple-subsidiary-schema.json
+- schema/schemas/subsequent-registration-schema.json
 - schema/schemas/subsidiary-relationship-schema.json
+- schema/schemas/supplier-relationship-schema.json
 - schema/schemas/trademark-registration-schema.json
 - spec/bots/bot-that-crashes-immediately/manifest.json
 - spec/bots/bot-that-crashes-immediately/scraper.rb
@@ -215,7 +223,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.5.1
+rubygems_version: 2.5.2.3
 signing_key: 
 specification_version: 4
 summary: Utilities for running bots with Turbot