rspec-hive 0.1.0 → 0.2.0

data/lib/rspec/hive/query_builder/null_strategy.rb

@@ -0,0 +1,11 @@
+module RSpec
+  module Hive
+    class QueryBuilder
+      class NullStrategy
+        def missing(_column)
+          '\N'
+        end
+      end
+    end
+  end
+end

data/lib/rspec/hive/query_builder/row_transformer.rb

@@ -0,0 +1,63 @@
+require_relative 'type_faker'
+
+module RSpec
+  module Hive
+    class QueryBuilder
+      class RowTransformer
+        def initialize(schema, missing_column_strategy)
+          @schema = schema
+          @strategy = missing_column_strategy
+        end
+
+        def transform(row)
+          if row.respond_to?(:each_pair)
+            mock_hive_row(row)
+          elsif row.respond_to?(:each)
+            array_row(row)
+          else
+            raise ArgumentError, 'Array or Hash required!'
+          end
+        end
+
+        private
+
+        attr_reader :schema, :strategy
+
+        HIVE_NIL = '\N'.freeze
+
+        def array_row(row)
+          size = schema.columns.size
+          missing = size - row.size
+          if missing > 0
+            row_with_missing_columns(row)
+          else
+            row
+          end
+        end
+
+        def row_with_missing_columns(row)
+          schema.columns.map.with_index do |column, index|
+            if index > row.size
+              strategy.missing(column)
+            else
+              row[index]
+            end
+          end
+        end
+
+        def mock_hive_row(partial_row)
+          symbolized_row = Hash[partial_row.map { |k, v| [k.to_sym, v] }]
+
+          schema.columns.map do |column|
+            value = symbolized_row.fetch(column.name.to_sym) { strategy.missing(column) }
+            nil_to_null(value)
+          end
+        end
+
+        def nil_to_null(value)
+          value.nil? ? HIVE_NIL : value
+        end
+      end
+    end
+  end
+end

data/lib/rspec/hive/query_builder/type_faker.rb

@@ -0,0 +1,36 @@
+require 'faker'
+
+module RSpec
+  module Hive
+    class QueryBuilder
+      class TypeFaker
+        class << self
+          # rubocop:disable Metrics/CyclomaticComplexity
+          def fake(type)
+            case type
+            when :int
+              Faker::Number.number(9)
+            when :smallint
+              Faker::Number.number(4)
+            when :tinyint
+              Faker::Number.number(1)
+            when :bigint
+              Faker::Number.number(12)
+            when :float
+              Faker::Number.decimal(4, 4)
+            when :double
+              Faker::Number.decimal(8, 8)
+            when :boolean
+              Faker::Boolean.boolean
+            when :string
+              Faker::Lorem.word
+            else
+              raise ArgumentError, "Unsupported type: #{type}"
+            end
+          end
+          # rubocop:enable Metrics/CyclomaticComplexity
+        end
+      end
+    end
+  end
+end

data/lib/rspec/hive/query_builder/value_by_type_strategy.rb

@@ -0,0 +1,13 @@
+require_relative 'type_faker'
+
+module RSpec
+  module Hive
+    class QueryBuilder
+      class ValueByTypeStrategy
+        def missing(column)
+          RSpec::Hive::QueryBuilder::TypeFaker.fake(column.type)
+        end
+      end
+    end
+  end
+end

data/lib/rspec/hive/query_builder_helper.rb

@@ -0,0 +1,24 @@
+require_relative 'query_builder'
+
+module RSpec
+  module Hive
+    module QueryBuilderHelper
+      def into_hive(schema)
+        hive_connection_guard!
+        ::RSpec::Hive::QueryBuilder.new(schema, connection)
+      end
+
+      private
+
+      def hive_connection_present?
+        respond_to?(:connection) &&
+          (connection.is_a?(RBHive::TCLIConnection) ||
+            connection.is_a?(RSpec::Hive::ConnectionDelegator))
+      end
+
+      def hive_connection_guard!
+        raise 'Include WithHiveConnection' unless hive_connection_present?
+      end
+    end
+  end
+end

data/lib/rspec/{rake_tasks → hive}/railtie.rb

@@ -2,9 +2,9 @@ require 'rails'
 
 module RSpec
   module Hive
-    class Railtie < Rails::Railtie
+    class Railtie < ::Rails::Railtie
       rake_tasks do
-        load 'rspec/rake_tasks/docker.rake'
+        load 'rspec/hive/rake_tasks/docker.rake'
       end
     end
   end

data/lib/rspec/hive/version.rb

@@ -1,5 +1,5 @@
 module RSpec
   module Hive
-    VERSION = '0.1.0'
+    VERSION = '0.2.0'.freeze
   end
 end

data/lib/rspec/hive/with_hive_connection.rb

@@ -1,17 +1,13 @@
+require_relative 'exponential_backoff'
+
 module RSpec
   module Hive
     module WithHiveConnection
-      def hive
-        Hive.connector
-      end
-
-      def connection
-        @connection ||= hive.start_connection
-      end
-
       def self.included(mod)
         mod.before(:all) do
-          connection
+          ExponentialBackoff.retryable(on: Thrift::TransportException) do
+            connection
+          end
         end
 
         mod.before(:each) do
@@ -19,9 +15,19 @@ module RSpec
         end
 
         mod.after(:all) do
-          hive.stop_connection(connection) unless hive && connection
+          hive_connector.stop_connection(connection) if hive_connector && @connection
         end
       end
+
+      def connection
+        @connection ||= hive_connector.start_connection
+      end
+
+      private
+
+      def hive_connector
+        ::RSpec::Hive.connector
+      end
     end
   end
 end

data/rspec-hive.gemspec

@@ -21,13 +21,15 @@ Gem::Specification.new do |spec|
 
   spec.add_dependency 'rake', '~> 10.0'
   spec.add_dependency 'colorize', '~> 0.7'
+  spec.add_dependency 'faker', '~> 1.6'
+  spec.add_dependency 'retryable', '~> 2.0.3'
+  spec.add_dependency 'rspec', '~> 3.4'
+  spec.add_dependency 'rbhive-u2i', '~> 1.0.0'
 
   spec.add_development_dependency 'bundler', '~> 1.7'
-  spec.add_development_dependency 'rspec', '~> 3.4'
   spec.add_development_dependency 'rspec-its', '~> 1.2'
-  spec.add_development_dependency 'rbhive', '~> 0.6.0'
-  spec.add_development_dependency 'rubocop', '~> 0.34'
-  spec.add_development_dependency 'rubocop-rspec', '~> 1.3'
+  spec.add_development_dependency 'rubocop', '~> 0.39'
+  spec.add_development_dependency 'rubocop-rspec', '~> 1.4'
   spec.add_development_dependency 'guard', '~> 2.6'
   spec.add_development_dependency 'guard-rspec', '~> 4.3'
   spec.add_development_dependency 'guard-rubocop', '~> 1.2'

data/spec/.rubocop.yml

@@ -0,0 +1,4 @@
+inherit_from: "../.rubocop_u2i.yml"
+
+Metrics/LineLength:
+  Enabled: false

data/spec/lib/rspec/hive/configuration_spec.rb

@@ -1,7 +1,7 @@
 require 'spec_helper'
 require 'tempfile'
 
-describe RSpec::Hive::Configuration do
+RSpec.describe RSpec::Hive::Configuration do
   RSpec.shared_examples('config') do
     its(:host) do
       is_expected.to eq(expected_host)
@@ -53,20 +53,21 @@ describe RSpec::Hive::Configuration do
 
   context 'when no configuration file is provided' do
     let(:expected_port) { 10000 }
+    let!(:original_host_os) { RbConfig::CONFIG['host_os'] }
 
     before { allow(Dir).to receive(:mktmpdir) { mock_tmpdir } }
 
-    subject { described_class.new }
-
     context 'when on Mac' do
       let(:mock_tmpdir) { '/Users/Shared/test/' }
       let(:expected_host) { '192.168.99.100' }
       let(:expected_host_shared_directory_path) { '/Users/Shared/test/spec-tmp-files' }
 
       before do
-        allow_any_instance_of(described_class).to receive(:mac?) { true }
+        RbConfig::CONFIG['host_os'] = 'mac os'
       end
 
+      after { RbConfig::CONFIG['host_os'] = original_host_os }
+
       include_examples('config')
     end
 
@@ -76,9 +77,11 @@ describe RSpec::Hive::Configuration do
       let(:expected_host_shared_directory_path) { '/tmp/test/spec-tmp-files' }
 
       before do
-        allow_any_instance_of(described_class).to receive(:mac?) { false }
+        RbConfig::CONFIG['host_os'] = 'linux'
       end
 
+      after { RbConfig::CONFIG['host_os'] = original_host_os }
+
       include_examples('config')
     end
   end
@@ -136,7 +139,7 @@ describe RSpec::Hive::Configuration do
       include_examples('config')
     end
 
-    context 'where there are some parametres required and optional' do
+    context 'where there are some parameters required and optional' do
       let(:yaml_hash) do
         {
           'hive' =>

data/spec/lib/rspec/hive/connection_delegator_spec.rb

@@ -1,6 +1,6 @@
 require 'spec_helper'
 
-describe RSpec::Hive::ConnectionDelegator do
+RSpec.describe RSpec::Hive::ConnectionDelegator do
   describe '#load_into_table' do
     let(:host_shared_directory_path) { '/tmp/host' }
     let(:docker_file_path) { '/tmp/docked/test_file' }

data/spec/lib/rspec/hive/connector_spec.rb

@@ -1,6 +1,6 @@
 require 'spec_helper'
 
-describe RSpec::Hive::Connector do
+RSpec.describe RSpec::Hive::Connector do
   describe '#start_connection' do
     let(:tcli_connection) { double(RBHive::TCLIConnection) }
     let(:connection_delegator) { double(RSpec::Hive::ConnectionDelegator) }

data/spec/lib/rspec/hive/db_name_spec.rb

@@ -1,6 +1,6 @@
 require 'spec_helper'
 
-describe RSpec::Hive::DbName do
+RSpec.describe RSpec::Hive::DbName do
   describe '.random_name' do
     subject { described_class.random_name }
 

data/spec/lib/rspec/hive/matchers_spec.rb

@@ -0,0 +1,94 @@
+require 'spec_helper'
+require 'rspec/hive/query_builder_helper'
+
+RSpec.describe 'match_result_set' do
+  include RSpec::Hive::QueryBuilderHelper
+
+  let(:john) { {first_name: 'John', last_name: 'Lennon', age: 40} }
+  let(:paul) { {first_name: 'Paul', last_name: 'McCartney', age: 73} }
+
+  let(:full_match) { expect(actual_rows).to match_result_set(expected_rows) }
+  let(:partial_match) { expect(actual_rows).to match_result_set(expected_rows).partially }
+  let(:full_match_fails) { expect(actual_rows).not_to match_result_set(expected_rows) }
+  let(:partial_match_fails) { expect(actual_rows).not_to match_result_set(expected_rows).partially }
+
+  context 'when the expected set has only one row' do
+    context 'but the actual set has more rows' do
+      let(:actual_rows) { [john, paul] }
+
+      context 'when the row is given as an array' do
+        let(:expected_rows) { [john.values] }
+
+        specify { full_match_fails }
+        specify { partial_match_fails }
+      end
+
+      context 'when the row is given as a hash' do
+        let(:expected_rows) { [john] }
+
+        specify { full_match_fails }
+        specify { partial_match_fails }
+      end
+    end
+
+    context 'and the actual set has one row' do
+      let(:actual_rows) { [john] }
+
+      context 'when the row is given as an array' do
+        context 'when the number of columns differs' do
+          let(:expected_rows) { [john.values << 'yoko'] }
+
+          specify { full_match_fails }
+        end
+
+        context 'when the actual and expected have are different' do
+          let(:expected_rows) { [paul.values] }
+
+          specify { full_match_fails }
+        end
+
+        context 'when the actual and expected rows are equal' do
+          let(:expected_rows) { [john.values] }
+
+          specify { full_match }
+        end
+
+        context 'when the actual and expected rows are equal with rspec matchers' do
+          let(:expected_rows) { [[a_string_matching('John'), a_string_matching(/lennon/i), 40]] }
+
+          specify { full_match }
+        end
+      end
+
+      context 'when the row is given as a hash' do
+        context 'when the number of columns differs' do
+          let(:expected_rows) { [john.dup.tap { |j| j[:ono] = 'yoko' }] }
+
+          specify { full_match_fails }
+          specify { partial_match_fails }
+        end
+
+        context 'when the actual and expected have are different' do
+          let(:expected_rows) { [john.dup.tap { |j| j[:first_name] = 'yoko' }] }
+
+          specify { full_match_fails }
+          specify { partial_match_fails }
+        end
+
+        context 'when the actual and expected rows are equal' do
+          let(:expected_rows) { [john] }
+
+          specify { full_match }
+          specify { partial_match }
+        end
+
+        context 'when matching a subset of columns' do
+          let(:expected_rows) { [{first_name: john[:first_name]}] }
+
+          specify { full_match_fails }
+          specify { partial_match }
+        end
+      end
+    end
+  end
+end

data/spec/lib/rspec/hive/query_builder/row_transformer_spec.rb

@@ -0,0 +1,37 @@
+require 'spec_helper'
+
+RSpec.describe RSpec::Hive::QueryBuilder::RowTransformer do
+  let(:transformer) { described_class.new(schema, missing_column_strategy) }
+  let(:schema) do
+    RBHive::TableSchema.new('table_name', nil) do
+      column :col1, :string
+      column :col2, :string
+    end
+  end
+  let(:column) { schema.instance_variable_get(:@columns).last }
+  let(:partition) { double }
+  let(:missing_column_strategy) { double }
+
+  describe '#transform' do
+    subject { transformer.transform(row) }
+
+    let(:row) { {col1: real_value} }
+    let(:real_value) { 'col1' }
+    let(:fake_value) { 'lorem' }
+    let(:expected_row) { [real_value, fake_value] }
+
+    before { allow(missing_column_strategy).to receive(:missing).with(column).and_return(fake_value) }
+
+    it 'fills missing fields' do
+      expect(subject[1]).to eq(fake_value)
+    end
+
+    it 'uses defined fields' do
+      expect(subject[0]).to eq(real_value)
+    end
+
+    it 'returns valid Rows' do
+      expect(subject).to eq(expected_row)
+    end
+  end
+end