csv_model 0.1.0 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 670a14423c708636825d5a2cebed117a0ac186e2
-  data.tar.gz: 7c39766c75b1b629c743036a5791da36c93032c9
+  metadata.gz: ccb22507d47dc2e7db940efdbb20d6366de3fb94
+  data.tar.gz: b7c9bf17cfac53e97271d63f6d3ea78634b2f8f6
 SHA512:
-  metadata.gz: 8fb5608b94c655f60c36c89b469f651ddd4a725a64fa5359b6534ec4cc37197cb80387e3bc213d40cb09b49569a0417ebfa63c60ccf4e81e54d2db41d71f46e4
-  data.tar.gz: 03c07a434abc2fec5416f62cff6453a4bffd2758f017e5c91cadb35daeca408adf26c03b1592e721ee6051c9be864f74f04e6d05c2c062a8a28bfffada80bc54
+  metadata.gz: 05ee82ff370463e138682ac77425d21e3dff73b6a876c4588a681a0efe45069936ce0b03d823f75ef086d0334ffd99cba8a80706b58341ed2336bba0a6f6492a
+  data.tar.gz: 31679c8cc0fc3314c9b27512940e1a2590997245791a41f33173c63887ea04d5a26b20c391621e795772130e81569676511dba30242dc8345365c193b8000cd9

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    csv_model (0.1.0)
+    csv_model (0.2.0)
 
 GEM
   remote: https://rubygems.org/

data/README.md

@@ -4,3 +4,7 @@ csv_model
 [![Build Status](https://travis-ci.org/Scrimmage/csv_model.svg)](https://travis-ci.org/Scrimmage/csv_model)
 
 [![Code Climate](https://codeclimate.com/github/Scrimmage/csv_model.png)](https://codeclimate.com/github/Scrimmage/csv_model)
+
+TODO: ability to have multiple alternate primary keys
+
+active_record?

data/lib/csv_model/extensions.rb

@@ -31,7 +31,7 @@ module CSVModel
       end
 
       def underscore
-        tr(' ', '_').tr("-", "_")        
+        tr(' ', '_').tr("-", "_")
       end
     end
 

data/lib/csv_model/header_row.rb

@@ -9,6 +9,7 @@ module CSVModel
     def initialize(data, options = {})
       @data = data
       @options = options
+      validate_options
     end
 
     def columns
@@ -24,24 +25,43 @@ module CSVModel
     end
 
     def errors
-      duplicate_column_errors + illegal_column_errors + missing_column_errors
+      duplicate_column_errors + illegal_column_errors + missing_column_errors + missing_key_column_errors
     end
 
-    # TODO: Remove? Not currently used.
     def has_column?(key)
       !column_index(key).nil?
     end
 
     def primary_key_columns
-      @primary_key_columns ||= columns.select { |x| primary_key_column_keys.include?(x.key) }
+      @primary_key_columns ||= begin
+        if has_primary_key? && has_primary_key_columns?
+          primary_primary_key_columns
+        elsif has_alternate_primary_key? && has_alternate_primary_key_columns?
+          alternate_primary_key_columns
+        else
+          []
+        end
+      end
     end
 
     def valid?
-      has_required_columns? && !has_duplicate_columns? && !has_illegal_columns?
+      has_required_columns? && has_required_key_columns? && !has_duplicate_columns? && !has_illegal_columns?
     end
 
     protected
 
+    def alternate_primary_key_columns
+      columns.select { |x| alternate_primary_key_column_keys.include?(x.key) }
+    end
+
+    def alternate_primary_key_column_keys
+      alternate_primary_key_column_names.collect { |x| x.to_column_key }
+    end
+
+    def alternate_primary_key_column_names
+      option(:alternate_primary_key, [])
+    end
+
     def column_keys
       @column_keys ||= columns.collect { |x| x.key }
     end
@@ -65,6 +85,14 @@ module CSVModel
         .collect { |key, count| column_name(key) }
     end
 
+    def has_alternate_primary_key?
+      alternate_primary_key_column_names.any?
+    end
+
+    def has_alternate_primary_key_columns?
+      missing_alternate_primary_key_column_keys.empty?
+    end
+
     def has_duplicate_columns?
       data.count != column_map.keys.count
     end
@@ -77,6 +105,18 @@ module CSVModel
       missing_column_keys.empty?
     end
 
+    def has_required_key_columns?
+      !has_primary_key? || (has_primary_key? && has_primary_key_columns?) || (has_alternate_primary_key? && has_alternate_primary_key_columns?)
+    end
+
+    def has_primary_key?
+      primary_key_column_names.any?
+    end
+
+    def has_primary_key_columns?
+      missing_primary_key_column_keys.empty?
+    end
+
     def illegal_column_errors
       illegal_column_names.collect { |name| "Unknown column #{name}" }
     end
@@ -97,6 +137,10 @@ module CSVModel
       legal_column_names.collect { |x| x.to_column_key }
     end
 
+    def missing_alternate_primary_key_column_keys
+      alternate_primary_key_column_keys - column_keys
+    end
+
     def missing_column_errors
       missing_column_names.collect { |name| "Missing column #{name}" }
     end
@@ -109,10 +153,34 @@ module CSVModel
       missing_column_keys.collect { |key| required_column_name(key) }
     end
 
+    def missing_primary_key_column_keys
+      primary_key_column_keys - column_keys
+    end
+
+    def missing_primary_key_column_names
+      missing_primary_key_column_keys.collect { |key| primary_key_column_name(key) }
+    end
+
+    def missing_key_column_errors
+      has_alternate_primary_key? && has_alternate_primary_key_columns? ? [] : primary_key_column_errors
+    end
+
+    def primary_primary_key_columns
+      columns.select { |x| primary_key_column_keys.include?(x.key) }
+    end
+
+    def primary_key_column_errors
+      missing_primary_key_column_names.collect { |name| "Missing primary key column #{name}" }
+    end
+
     def primary_key_column_keys
       primary_key_column_names.collect { |x| x.to_column_key }
     end
 
+    def primary_key_column_name(column_key)
+      primary_key_column_names.find { |entry| entry.to_column_key == column_key }
+    end
+
     def primary_key_column_names
       option(:primary_key, [])
     end
@@ -129,5 +197,27 @@ module CSVModel
       option(:required_columns, [])
     end
 
+    def validate_options
+      if !option(:primary_key).nil? && primary_key_column_keys.empty?
+        raise ArgumentError.new("The primary_key cannot be be empty.")
+      end
+
+      if legal_column_keys.any? && (primary_key_column_keys - legal_column_keys).any?
+        raise ArgumentError.new("The primary_key cannot contain columns that are not included in legal_column_keys.")
+      end
+
+      if primary_key_column_names.empty? && alternate_primary_key_column_names.any?
+        raise ArgumentError.new("The alternate_primary_key cannot be specified if no primary_key is specified.")
+      end
+
+      if !option(:alternate_primary_key).nil? && primary_key_column_keys == alternate_primary_key_column_keys
+        raise ArgumentError.new("The alternate_primary_key cannot be identical to the primary_key.")
+      end
+
+      if legal_column_keys.any? && (alternate_primary_key_column_keys - legal_column_keys).any?
+        raise ArgumentError.new("The alternate_primary_key cannot contain columns that are not included in legal_column_keys.")
+      end
+    end
   end
+
 end

data/lib/csv_model/model.rb

@@ -47,12 +47,16 @@ module CSVModel
       row
     end
 
+    def detect_duplicate_rows?
+      option(:detect_duplicate_rows, true)
+    end
+
     def header_class
       option(:header_class, HeaderRow)
     end
 
     def is_duplicate_key?(value)
-      return false if value.nil? || value == "" || (value.is_a?(Array) && value.all_values_blank?)
+      return false if !detect_duplicate_rows? || value.nil? || value == "" || (value.is_a?(Array) && value.all_values_blank?)
       !keys.add?(value)
     end
 

data/lib/csv_model/utilities.rb

@@ -7,7 +7,10 @@ module CSVModel
       attr_reader :options
 
       def option(key, default = nil)
-        options.try(:[], key) || options.try(key) || default
+        value = options.try(:[], key) 
+        value = options.try(key) if value.nil?
+        value = default if value.nil?
+        value
       end
 
     end

data/lib/csv_model/version.rb

@@ -1,3 +1,3 @@
 module CSVModel
-  VERSION = "0.1.0"
+  VERSION = "0.2.0"
 end

data/spec/csv_model/header_row_spec.rb

@@ -5,9 +5,31 @@ describe CSVModel::HeaderRow do
   let(:data) { ["Column One"] }
   let(:subject) { described_class.new(data) }
 
-  # TODO: should detect invalid opbions
-  #  e.g. when primar_key is not a subset of columns
-  #   raise ArgumentError, 'Message'
+  describe "#new" do
+    it "raises an ArgumentError when primary_key is not a subset of legal columns" do
+      expect { described_class.new(data, legal_columns: ["Column One"], primary_key: ["Column Two"]) }.to raise_error(ArgumentError)
+    end
+
+    it "does not raises an ArgumentError when primary_key is a subset of legal columns" do
+      expect { described_class.new(data, legal_columns: ["Column One"], primary_key: ["column one"]) }.to_not raise_error
+    end
+
+    it "raises an ArgumentError when primary_key is specified but empty" do
+      expect { described_class.new(data, primary_key: []) }.to raise_error(ArgumentError)
+    end
+
+    it "raises an ArgumentError when alternative_primary_key is specified but no primary_key is specified" do
+      expect { described_class.new(data, alternate_primary_key: ["Column One"]) }.to raise_error(ArgumentError)
+    end
+
+    it "raises an ArgumentError when alternative_primary_key is not a subset of legal columns" do
+      expect { described_class.new(data, legal_columns: ["Column One"], primary_key: ["Column One"], alternate_primary_key: ["Column Two"]) }.to raise_error(ArgumentError)
+    end
+
+    it "raises an ArgumentError when alternative_primary_key colums are idential to primary key columns" do
+      expect { described_class.new(data, primary_key: ["Column One"], alternate_primary_key: ["Column One"]) }.to raise_error(ArgumentError)
+    end
+  end
 
   describe "#columns" do
     it "returns an empty array when no columns exist" do
@@ -77,10 +99,25 @@ describe CSVModel::HeaderRow do
       expect(subject.errors).to eq(["Unknown column Column One"])
     end
 
-    it "returns returns required column message when header is missing required columns" do
+    it "returns required column message when header is missing required columns" do
       subject = described_class.new(data, required_columns: ["Column Two"])
       expect(subject.errors).to eq(["Missing column Column Two"])
     end
+
+    it "returns required column message when header is missing required primary key columns" do
+      subject = described_class.new(data, primary_key: ["Column Two"])
+      expect(subject.errors).to eq(["Missing primary key column Column Two"])
+    end
+
+    it "returns empty array when header is missing required primary key column but alternate is available" do
+      subject = described_class.new(data, primary_key: ["Column Two"], alternate_primary_key: ["Column One"])
+      expect(subject.errors).to eq([])
+    end
+
+    it "returns required column message when header is missing required primary key and alternate is not available " do
+      subject = described_class.new(data, primary_key: ["Column Two"], alternate_primary_key: ["Column Three"])
+      expect(subject.errors).to eq(["Missing primary key column Column Two"])
+    end
   end
 
   describe "#has_column?" do
@@ -118,6 +155,15 @@ describe CSVModel::HeaderRow do
           expect(column.name).to eq("Column One")
         end
       end
+
+      context "when the primary key column is not present but an alternate is available and present" do
+        let(:subject) { described_class.new(data, OpenStruct.new(primary_key: ["column four"], alternate_primary_key: ["column one"])) }
+
+        it "returns the alternate key column" do
+          column = subject.primary_key_columns.first
+          expect(column.name).to eq("Column One")
+        end
+      end
     end
 
     context "when the row has multiple primary key columns" do
@@ -145,10 +191,25 @@ describe CSVModel::HeaderRow do
       expect(subject.valid?).to eq(false)
     end
 
+    it "returns false when header is missing a primary key column" do
+      subject = described_class.new(data, primary_key: ["Column Two"])
+      expect(subject.valid?).to eq(false)
+    end
+
     it "returns false when header is missing required columns" do
       subject = described_class.new(data, required_columns: ["Column Two"])
       expect(subject.valid?).to eq(false)
     end
+
+    it "returns true when header is missing required primary key column but alternate is available" do
+      subject = described_class.new(data, primary_key: ["Column Two"], alternate_primary_key: ["Column One"])
+      expect(subject.valid?).to eq(true)
+    end
+
+    it "returns false when header is missing required primary key and alternate is not available " do
+      subject = described_class.new(data, primary_key: ["Column Two"], alternate_primary_key: ["Column Three"])
+      expect(subject.valid?).to eq(false)
+    end
   end
 
   describe "internals" do
@@ -167,6 +228,47 @@ describe CSVModel::HeaderRow do
       end
     end
 
+    describe "#has_alternate_primary_key?" do
+      it "doesn't respond to has_alternate_primary_key?" do
+        expect(subject.respond_to?(:has_alternate_primary_key?)).to eq(false)
+      end
+
+      it "returns false when an alternate primary key is not specified" do
+        expect(subject.send(:has_alternate_primary_key?)).to eq(false)
+      end
+
+      context "with an alternate primary key" do
+        let(:subject) { described_class.new(data, primary_key: ["Column Four"], alternate_primary_key: ["Column One"]) }
+
+        it "returns true" do
+          expect(subject.send(:has_alternate_primary_key?)).to eq(true)
+        end
+      end
+    end
+
+    describe "#has_alternate_primary_key_columns?" do
+      it "doesn't respond to has_alternate_primary_key_columns?" do
+        expect(subject.respond_to?(:has_alternate_primary_key_columns?)).to eq(false)
+      end
+
+      it "returns true when no primary key columns specified" do        
+        expect(subject.send(:has_alternate_primary_key_columns?)).to eq(true)
+      end
+
+      context "with alternate key columns specified" do
+        let(:subject) { described_class.new(data, primary_key: ["Column Four"], alternate_primary_key: ["Column One"]) }
+
+        it "returns true when all alternate primary key columns are present" do
+          expect(subject.send(:has_alternate_primary_key_columns?)).to eq(true)
+        end
+
+        it "returns false when an alternate primry key column is missing" do
+          data[0] = "Column Two"
+          expect(subject.send(:has_alternate_primary_key_columns?)).to eq(false)
+        end
+      end
+    end
+
     describe "#has_duplicate_columns?" do
       it "doesn't respond to has_duplicate_columns?" do
         expect(subject.respond_to?(:has_duplicate_columns?)).to eq(false)
@@ -205,6 +307,47 @@ describe CSVModel::HeaderRow do
       end
     end
 
+    describe "#has_primary_key?" do
+      it "doesn't respond to has_primary_key?" do
+        expect(subject.respond_to?(:has_primary_key?)).to eq(false)
+      end
+
+      it "returns false when a primary key is not specified" do
+        expect(subject.send(:has_primary_key?)).to eq(false)
+      end
+
+      context "with a primary key" do
+        let(:subject) { described_class.new(data, primary_key: ["Column One"]) }
+
+        it "returns true" do
+          expect(subject.send(:has_primary_key?)).to eq(true)
+        end
+      end
+    end
+
+    describe "#has_primary_key_columns?" do
+      it "doesn't respond to has_primary_key_columns?" do
+        expect(subject.respond_to?(:has_primary_key_columns?)).to eq(false)
+      end
+
+      it "returns true when no primary key columns specified" do
+        expect(subject.send(:has_primary_key_columns?)).to eq(true)
+      end
+
+      context "with primary key columns specified" do
+        let(:subject) { described_class.new(data, primary_key: ["Column One"]) }
+
+        it "returns true when all primary key columns are present" do
+          expect(subject.send(:has_primary_key_columns?)).to eq(true)
+        end
+
+        it "returns false when a primry key column is missing" do
+          data[0] = "Column Two"
+          expect(subject.send(:has_primary_key_columns?)).to eq(false)
+        end
+      end
+    end
+
     describe "#has_required_columns?" do
       it "doesn't respond to has_required_columns?" do
         expect(subject.respond_to?(:has_required_columns?)).to eq(false)
@@ -228,7 +371,49 @@ describe CSVModel::HeaderRow do
       end
     end
 
-    describe "illegal_column_names" do
+    describe "#has_required_key_columns?" do
+      it "doesn't respond to has_required_key_columns?" do
+        expect(subject.respond_to?(:has_required_key_columns?)).to eq(false)
+      end
+
+      it "returns true when no primary key is specified" do
+        expect(subject.send(:has_required_key_columns?)).to eq(true)
+      end
+
+      context "with primary key columns specified" do
+        let(:subject) { described_class.new(data, primary_key: ["Column One"]) }
+
+        it "returns true when all primary key columns are present" do
+          expect(subject.send(:has_required_key_columns?)).to eq(true)
+        end
+
+        it "returns false when a primary key columns is missing" do
+          data[0] = "Column Two"
+          expect(subject.send(:has_required_key_columns?)).to eq(false)
+        end
+      end
+
+      context "with primary and alternate key columns specified" do
+        let(:subject) { described_class.new(data, primary_key: ["Column Two"], alternate_primary_key: ["Column Four"]) }
+
+        it "returns true when all primary key columns are present" do
+          data[0] = "Column Two"
+          expect(subject.send(:has_required_key_columns?)).to eq(true)
+        end
+
+        it "returns true when a primary key columns is missing but all alternate primary key columns are present" do
+          data[0] = "Column Four"
+          expect(subject.send(:has_required_key_columns?)).to eq(true)
+        end
+
+        it "returns false when a primary key column is missing and no alternative is available" do
+          data[0] = "Column Three"
+          expect(subject.send(:has_required_key_columns?)).to eq(false)
+        end
+      end
+    end
+
+    describe "#illegal_column_names" do
       it "doesn't respond to illegal_colum_names" do
         expect(subject.respond_to?(:illegal_colum_names)).to eq(false)
       end
@@ -273,6 +458,29 @@ describe CSVModel::HeaderRow do
         end
       end
     end
+
+    describe "#missing_primary_key_column_names" do
+      it "doesn't respond to missing_primary_key_column_names" do
+        expect(subject.respond_to?(:missing_primary_key_column_names)).to eq(false)
+      end
+
+      it "returns emptry array when no primary key columns specified" do
+        expect(subject.send(:missing_primary_key_column_names)).to eq([])
+      end
+
+      context "with primary key columns specified" do
+        let(:subject) { described_class.new(data, OpenStruct.new(primary_key: ["Column One"])) }
+
+        it "returns empty array when all primary key columns are present" do
+          expect(subject.send(:missing_primary_key_column_names)).to eq([])
+        end
+
+        it "returns array of missing column names when a primary key column is missing" do
+          data[0] = "Column Two"
+          expect(subject.send(:missing_primary_key_column_names)).to eq(["Column One"])
+        end
+      end
+    end
   end
 
 end

data/spec/csv_model/model_spec.rb

@@ -139,12 +139,19 @@ describe CSVModel::Model do
           subject.send(:parse_data)
         end
 
+        context "with duplicate checking disabled" do
+          let(:subject) { described_class.new(data, detect_duplicate_rows: false, primary_key: [header_row.first]) }
+
+          it "does not mark any row as a duplicate" do
+            subject.rows.each { |row| expect(row.marked_as_duplicate?).to eq(false) }
+          end
+        end
+
         context "with a single column primary key" do
           let(:subject) { described_class.new(data, primary_key: [header_row.first]) }
 
           context "when primary key values are present" do
             it "does not mark the first row as a duplicate" do
-              puts subject.structure_errors
               expect(subject.rows.first.marked_as_duplicate?).to eq(false)
             end
 
@@ -185,7 +192,6 @@ describe CSVModel::Model do
             end
           end
         end
-
       end
     end
   end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: csv_model
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.2.0
 platform: ruby
 authors:
 - Matthew Chadwick
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-07-01 00:00:00.000000000 Z
+date: 2014-07-11 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler